Mercurial > repos > eschen42 > mqppep_preproc

annotate mqppep_mrgfltr.py @ 22:43f1fd0ff86b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 00881af5b1373174e5afe706add2f33b8614828c"
author eschen42
date Wed, 13 Apr 2022 19:48:01 +0000
parents ba5f14c2a4af
children 7560a4e80a1e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1 #!/usr/bin/env python
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
2
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
3 # Import the packages needed
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
4 import argparse
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
5 import operator # for operator.itemgetter
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
6 import os.path
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
7 import re
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
8 import shutil # for shutil.copyfile(src, dest)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
9 import sqlite3 as sql
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
10 import sys # import the sys module for exc_info
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
11 import time
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
12 import traceback # for formatting stack-trace
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
13 from codecs import getreader as cx_getreader
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
14
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
15 import numpy as np
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
16 import pandas
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
17
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
18 # global constants
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
19 N_A = "N/A"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
20
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
21
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
22 # ref: https://stackoverflow.com/a/8915613/15509512
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
23 # answers: "How to handle exceptions in a list comprehensions"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
24 # usage:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
25 # from math import log
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
26 # eggs = [1,3,0,3,2]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
27 # print([x for x in [catch(log, egg) for egg in eggs] if x is not None])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
28 # producing:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
29 # for <built-in function log>
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
30 # with args (0,)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
31 # exception: math domain error
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
32 # [0.0, 1.0986122886681098, 1.0986122886681098, 0.6931471805599453]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
33 def catch(func, *args, handle=lambda e: e, **kwargs):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
34
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
35 try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
36 return func(*args, **kwargs)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
37 except Exception as e:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
38 print("For %s" % str(func))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
39 print(" with args %s" % str(args))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
40 print(" caught exception: %s" % str(e))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
41 (ty, va, tb) = sys.exc_info()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
42 print(" stack trace: " + str(traceback.format_exception(ty, va, tb)))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
43 exit(-1)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
44 return None
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
45
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
46
6
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
47 def whine(func, *args, handle=lambda e: e, **kwargs):
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
48
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
49 try:
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
50 return func(*args, **kwargs)
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
51 except Exception as e:
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
52 print("Warning: For %s" % str(func))
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
53 print(" with args %s" % str(args))
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
54 print(" caught exception: %s" % str(e))
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
55 (ty, va, tb) = sys.exc_info()
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
56 print(" stack trace: " + str(traceback.format_exception(ty, va, tb)))
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
57 return None
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
58
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
59
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
60 def ppep_join(x):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
61 x = [i for i in x if N_A != i]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
62 result = "%s" % " | ".join(x)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
63 if result != "":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
64 return result
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
65 else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
66 return N_A
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
67
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
68
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
69 def melt_join(x):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
70 tmp = {key.lower(): key for key in x}
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
71 result = "%s" % " | ".join([tmp[key] for key in tmp])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
72 return result
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
73
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
74
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
75 def __main__():
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
76 # Parse Command Line
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
77 parser = argparse.ArgumentParser(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
78 description="Phopsphoproteomic Enrichment Pipeline Merge and Filter."
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
79 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
80
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
81 # inputs:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
82 # Phosphopeptide data for experimental results, including the intensities
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
83 # and the mapping to kinase domains, in tabular format.
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
84 parser.add_argument(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
85 "--phosphopeptides",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
86 "-p",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
87 nargs=1,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
88 required=True,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
89 dest="phosphopeptides",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
90 help="Phosphopeptide data for experimental results, including the intensities and the mapping to kinase domains, in tabular format",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
91 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
92 # UniProtKB/SwissProt DB input, SQLite
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
93 parser.add_argument(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
94 "--ppep_mapping_db",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
95 "-d",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
96 nargs=1,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
97 required=True,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
98 dest="ppep_mapping_db",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
99 help="UniProtKB/SwissProt SQLite Database",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
100 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
101 # species to limit records chosed from PhosPhositesPlus
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
102 parser.add_argument(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
103 "--species",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
104 "-x",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
105 nargs=1,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
106 required=False,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
107 default=[],
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
108 dest="species",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
109 help="limit PhosphoSitePlus records to indicated species (field may be empty)",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
110 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
111
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
112 # outputs:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
113 # tabular output
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
114 parser.add_argument(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
115 "--mrgfltr_tab",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
116 "-o",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
117 nargs=1,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
118 required=True,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
119 dest="mrgfltr_tab",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
120 help="Tabular output file for results",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
121 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
122 # CSV output
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
123 parser.add_argument(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
124 "--mrgfltr_csv",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
125 "-c",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
126 nargs=1,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
127 required=True,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
128 dest="mrgfltr_csv",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
129 help="CSV output file for results",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
130 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
131 # SQLite output
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
132 parser.add_argument(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
133 "--mrgfltr_sqlite",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
134 "-S",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
135 nargs=1,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
136 required=True,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
137 dest="mrgfltr_sqlite",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
138 help="SQLite output file for results",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
139 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
140
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
141 # "Make it so!" (parse the arguments)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
142 options = parser.parse_args()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
143 print("options: " + str(options))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
144
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
145 # determine phosphopeptide ("upstream map") input tabular file access
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
146 if options.phosphopeptides is None:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
147 exit('Argument "phosphopeptides" is required but not supplied')
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
148 try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
149 upstream_map_filename_tab = os.path.abspath(options.phosphopeptides[0])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
150 input_file = open(upstream_map_filename_tab, "r")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
151 input_file.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
152 except Exception as e:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
153 exit("Error parsing phosphopeptides argument: %s" % str(e))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
154
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
155 # determine input SQLite access
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
156 if options.ppep_mapping_db is None:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
157 exit('Argument "ppep_mapping_db" is required but not supplied')
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
158 try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
159 uniprot_sqlite = os.path.abspath(options.ppep_mapping_db[0])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
160 input_file = open(uniprot_sqlite, "rb")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
161 input_file.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
162 except Exception as e:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
163 exit("Error parsing ppep_mapping_db argument: %s" % str(e))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
164
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
165 # copy input SQLite dataset to output SQLite dataset
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
166 if options.mrgfltr_sqlite is None:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
167 exit('Argument "mrgfltr_sqlite" is required but not supplied')
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
168 try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
169 output_sqlite = os.path.abspath(options.mrgfltr_sqlite[0])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
170 shutil.copyfile(uniprot_sqlite, output_sqlite)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
171 except Exception as e:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
172 exit("Error copying ppep_mapping_db to mrgfltr_sqlite: %s" % str(e))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
173
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
174 # determine species to limit records from PSP_Regulatory_Sites
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
175 if options.species is None:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
176 exit(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
177 'Argument "species" is required (and may be empty) but not supplied'
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
178 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
179 try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
180 if len(options.species) > 0:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
181 species = options.species[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
182 else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
183 species = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
184 except Exception as e:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
185 exit("Error parsing species argument: %s" % str(e))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
186
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
187 # determine tabular output destination
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
188 if options.mrgfltr_tab is None:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
189 exit('Argument "mrgfltr_tab" is required but not supplied')
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
190 try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
191 output_filename_tab = os.path.abspath(options.mrgfltr_tab[0])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
192 output_file = open(output_filename_tab, "w")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
193 output_file.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
194 except Exception as e:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
195 exit("Error parsing mrgfltr_tab argument: %s" % str(e))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
196
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
197 # determine CSV output destination
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
198 if options.mrgfltr_csv is None:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
199 exit('Argument "mrgfltr_csv" is required but not supplied')
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
200 try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
201 output_filename_csv = os.path.abspath(options.mrgfltr_csv[0])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
202 output_file = open(output_filename_csv, "w")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
203 output_file.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
204 except Exception as e:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
205 exit("Error parsing mrgfltr_csv argument: %s" % str(e))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
206
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
207 def mqpep_getswissprot():
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
208
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
209 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
210 # copied from Excel Output Script.ipynb BEGIN #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
211 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
212
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
213 # String Constants #################
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
214 DEPHOSPHOPEP = "DephosphoPep"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
215 DESCRIPTION = "Description"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
216 FUNCTION_PHOSPHORESIDUE = (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
217 "Function Phosphoresidue(PSP=PhosphoSitePlus.org)"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
218 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
219 GENE_NAME = "Gene_Name" # Gene Name from UniProtKB
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
220 ON_FUNCTION = (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
221 "ON_FUNCTION" # ON_FUNCTION column from PSP_Regulatory_Sites
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
222 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
223 ON_NOTES = "NOTES" # NOTES column from PSP_Regulatory_Sites
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
224 ON_OTHER_INTERACT = "ON_OTHER_INTERACT" # ON_OTHER_INTERACT column from PSP_Regulatory_Sites
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
225 ON_PROCESS = (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
226 "ON_PROCESS" # ON_PROCESS column from PSP_Regulatory_Sites
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
227 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
228 ON_PROT_INTERACT = "ON_PROT_INTERACT" # ON_PROT_INTERACT column from PSP_Regulatory_Sites
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
229 PHOSPHOPEPTIDE = "Phosphopeptide"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
230 PHOSPHOPEPTIDE_MATCH = "Phosphopeptide_match"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
231 PHOSPHORESIDUE = "Phosphoresidue"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
232 PUTATIVE_UPSTREAM_DOMAINS = "Putative Upstream Kinases(PSP=PhosphoSitePlus.org)/Phosphatases/Binding Domains"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
233 SEQUENCE = "Sequence"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
234 SEQUENCE10 = "Sequence10"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
235 SEQUENCE7 = "Sequence7"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
236 SITE_PLUSMINUS_7AA_SQL = "SITE_PLUSMINUS_7AA"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
237 UNIPROT_ID = "UniProt_ID"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
238 UNIPROT_SEQ_AND_META_SQL = """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
239 select Uniprot_ID, Description, Gene_Name, Sequence,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
240 Organism_Name, Organism_ID, PE, SV
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
241 from UniProtKB
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
242 order by Sequence, UniProt_ID
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
243 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
244 UNIPROT_UNIQUE_SEQ_SQL = """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
245 select distinct Sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
246 from UniProtKB
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
247 group by Sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
248 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
249 PPEP_PEP_UNIPROTSEQ_SQL = """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
250 select distinct phosphopeptide, peptide, sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
251 from uniprotkb_pep_ppep_view
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
252 order by sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
253 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
254 PPEP_MELT_SQL = """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
255 SELECT DISTINCT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
256 phospho_peptide AS 'p_peptide',
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
257 kinase_map AS 'characterization',
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
258 'X' AS 'X'
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
259 FROM ppep_gene_site_view
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
260 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
261 # CREATE TABLE PSP_Regulatory_site (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
262 # site_plusminus_7AA TEXT PRIMARY KEY ON CONFLICT IGNORE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
263 # domain TEXT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
264 # ON_FUNCTION TEXT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
265 # ON_PROCESS TEXT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
266 # ON_PROT_INTERACT TEXT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
267 # ON_OTHER_INTERACT TEXT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
268 # notes TEXT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
269 # organism TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
270 # );
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
271 PSP_REGSITE_SQL = """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
272 SELECT DISTINCT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
273 SITE_PLUSMINUS_7AA ,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
274 DOMAIN ,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
275 ON_FUNCTION ,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
276 ON_PROCESS ,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
277 ON_PROT_INTERACT ,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
278 ON_OTHER_INTERACT ,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
279 NOTES ,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
280 ORGANISM
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
281 FROM PSP_Regulatory_site
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
282 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
283 PPEP_ID_SQL = """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
284 SELECT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
285 id AS 'ppep_id',
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
286 seq AS 'ppep_seq'
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
287 FROM ppep
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
288 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
289 MRGFLTR_DDL = """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
290 DROP VIEW IF EXISTS mrgfltr_metadata_view;
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
291 DROP TABLE IF EXISTS mrgfltr_metadata;
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
292 CREATE TABLE mrgfltr_metadata
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
293 ( ppep_id INTEGER REFERENCES ppep(id)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
294 , Sequence10 TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
295 , Sequence7 TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
296 , GeneName TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
297 , Phosphoresidue TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
298 , UniProtID TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
299 , Description TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
300 , FunctionPhosphoresidue TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
301 , PutativeUpstreamDomains TEXT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
302 , PRIMARY KEY (ppep_id) ON CONFLICT IGNORE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
303 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
304 ;
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
305 CREATE VIEW mrgfltr_metadata_view AS
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
306 SELECT DISTINCT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
307 ppep.seq AS phospho_peptide
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
308 , Sequence10
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
309 , Sequence7
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
310 , GeneName
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
311 , Phosphoresidue
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
312 , UniProtID
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
313 , Description
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
314 , FunctionPhosphoresidue
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
315 , PutativeUpstreamDomains
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
316 FROM
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
317 ppep, mrgfltr_metadata
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
318 WHERE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
319 mrgfltr_metadata.ppep_id = ppep.id
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
320 ORDER BY
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
321 ppep.seq
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
322 ;
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
323 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
324
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
325 CITATION_INSERT_STMT = """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
326 INSERT INTO Citation (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
327 ObjectName,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
328 CitationData
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
329 ) VALUES (?,?)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
330 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
331 CITATION_INSERT_PSP = 'PhosphoSitePlus(R) (PSP) was created by Cell Signaling Technology Inc. It is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. When using PSP data or analyses in printed publications or in online resources, the following acknowledgements must be included: (a) the words "PhosphoSitePlus(R), www.phosphosite.org" must be included at appropriate places in the text or webpage, and (b) the following citation must be included in the bibliography: "Hornbeck PV, Zhang B, Murray B, Kornhauser JM, Latham V, Skrzypek E PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res. 2015 43:D512-20. PMID: 25514926."'
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
332 CITATION_INSERT_PSP_REF = 'Hornbeck, 2014, "PhosphoSitePlus, 2014: mutations, PTMs and recalibrations.", https://pubmed.ncbi.nlm.nih.gov/22135298, https://doi.org/10.1093/nar/gkr1122'
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
333
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
334 MRGFLTR_METADATA_COLUMNS = [
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
335 "ppep_id",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
336 "Sequence10",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
337 "Sequence7",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
338 "GeneName",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
339 "Phosphoresidue",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
340 "UniProtID",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
341 "Description",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
342 "FunctionPhosphoresidue",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
343 "PutativeUpstreamDomains",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
344 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
345
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
346 # String Constants (end) ############
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
347
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
348 class Error(Exception):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
349 """Base class for exceptions in this module."""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
350
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
351 pass
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
352
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
353 class PreconditionError(Error):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
354 """Exception raised for errors in the input.
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
355
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
356 Attributes:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
357 expression -- input expression in which the error occurred
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
358 message -- explanation of the error
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
359 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
360
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
361 def __init__(self, expression, message):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
362 self.expression = expression
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
363 self.message = message
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
364
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
365 # start_time = time.clock() #timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
366 start_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
367
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
368 # get keys from upstream tabular file using readline()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
369 # ref: https://stackoverflow.com/a/16713581/15509512
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
370 # answer to "Use codecs to read file with correct encoding"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
371 file1_encoded = open(upstream_map_filename_tab, "rb")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
372 file1 = cx_getreader("latin-1")(file1_encoded)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
373
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
374 count = 0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
375 upstream_map_p_peptide_list = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
376 re_tab = re.compile("^[^\t]*")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
377 while True:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
378 count += 1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
379 # Get next line from file
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
380 line = file1.readline()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
381 # if line is empty
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
382 # end of file is reached
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
383 if not line:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
384 break
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
385 if count > 1:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
386 m = re_tab.match(line)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
387 upstream_map_p_peptide_list.append(m[0])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
388 file1.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
389 file1_encoded.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
390
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
391 # Get the list of phosphopeptides with the p's that represent the phosphorylation sites removed
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
392 re_phos = re.compile("p")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
393
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
394 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
395 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
396 "%0.6f pre-read-SwissProt [0.1]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
397 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
398 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
399
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
400 # ----------- Get SwissProt data from SQLite database (start) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
401 # build UniProt sequence LUT and list of unique SwissProt sequences
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
402
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
403 # Open SwissProt SQLite database
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
404 conn = sql.connect(uniprot_sqlite)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
405 cur = conn.cursor()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
406
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
407 # Set up structures to hold SwissProt data
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
408
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
409 uniprot_Sequence_List = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
410 UniProtSeqLUT = {}
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
411
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
412 # Execute query for unique seqs without fetching the results yet
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
413 uniprot_unique_seq_cur = cur.execute(UNIPROT_UNIQUE_SEQ_SQL)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
414
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
415 while 1:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
416 batch = uniprot_unique_seq_cur.fetchmany(size=50)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
417 if not batch:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
418 # handle case where no records are returned
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
419 break
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
420 for row in batch:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
421 Sequence = row[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
422 UniProtSeqLUT[(Sequence, DESCRIPTION)] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
423 UniProtSeqLUT[(Sequence, GENE_NAME)] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
424 UniProtSeqLUT[(Sequence, UNIPROT_ID)] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
425 UniProtSeqLUT[Sequence] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
426
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
427 # Execute query for seqs and metadata without fetching the results yet
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
428 uniprot_seq_and_meta = cur.execute(UNIPROT_SEQ_AND_META_SQL)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
429
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
430 while 1:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
431 batch = uniprot_seq_and_meta.fetchmany(size=50)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
432 if not batch:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
433 # handle case where no records are returned
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
434 break
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
435 for (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
436 UniProt_ID,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
437 Description,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
438 Gene_Name,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
439 Sequence,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
440 OS,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
441 OX,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
442 PE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
443 SV,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
444 ) in batch:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
445 uniprot_Sequence_List.append(Sequence)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
446 UniProtSeqLUT[Sequence] = Sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
447 UniProtSeqLUT[(Sequence, UNIPROT_ID)].append(UniProt_ID)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
448 UniProtSeqLUT[(Sequence, GENE_NAME)].append(Gene_Name)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
449 if OS != N_A:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
450 Description += " OS=" + OS
17
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 16
diff changeset
451 if OX != -1:
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 16
diff changeset
452 Description += " OX=" + str(OX)
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
453 if Gene_Name != N_A:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
454 Description += " GN=" + Gene_Name
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
455 if PE != N_A:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
456 Description += " PE=" + PE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
457 if SV != N_A:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
458 Description += " SV=" + SV
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
459 UniProtSeqLUT[(Sequence, DESCRIPTION)].append(Description)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
460
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
461 # Close SwissProt SQLite database; clean up local variables
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
462 conn.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
463 Sequence = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
464 UniProt_ID = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
465 Description = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
466 Gene_Name = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
467
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
468 # ----------- Get SwissProt data from SQLite database (finish) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
469
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
470 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
471 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
472 "%0.6f post-read-SwissProt [0.2]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
473 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
474 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
475
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
476 # ----------- Get SwissProt data from SQLite database (start) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
477 # Open SwissProt SQLite database
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
478 conn = sql.connect(uniprot_sqlite)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
479 cur = conn.cursor()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
480
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
481 # Set up dictionary to aggregate results for phosphopeptides correspounding to dephosphoeptide
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
482 DephosphoPep_UniProtSeq_LUT = {}
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
483
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
484 # Set up dictionary to accumulate results
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
485 PhosphoPep_UniProtSeq_LUT = {}
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
486
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
487 # Execute query for tuples without fetching the results yet
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
488 ppep_pep_uniprotseq_cur = cur.execute(PPEP_PEP_UNIPROTSEQ_SQL)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
489
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
490 while 1:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
491 batch = ppep_pep_uniprotseq_cur.fetchmany(size=50)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
492 if not batch:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
493 # handle case where no records are returned
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
494 break
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
495 for (phospho_pep, dephospho_pep, sequence) in batch:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
496 # do interesting stuff here...
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
497 PhosphoPep_UniProtSeq_LUT[phospho_pep] = phospho_pep
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
498 PhosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
499 (phospho_pep, DEPHOSPHOPEP)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
500 ] = dephospho_pep
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
501 if dephospho_pep not in DephosphoPep_UniProtSeq_LUT:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
502 DephosphoPep_UniProtSeq_LUT[dephospho_pep] = set()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
503 DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
504 (dephospho_pep, DESCRIPTION)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
505 ] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
506 DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
507 (dephospho_pep, GENE_NAME)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
508 ] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
509 DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
510 (dephospho_pep, UNIPROT_ID)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
511 ] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
512 DephosphoPep_UniProtSeq_LUT[(dephospho_pep, SEQUENCE)] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
513 DephosphoPep_UniProtSeq_LUT[dephospho_pep].add(phospho_pep)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
514
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
515 if (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
516 sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
517 not in DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
518 (dephospho_pep, SEQUENCE)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
519 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
520 ):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
521 DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
522 (dephospho_pep, SEQUENCE)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
523 ].append(sequence)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
524 for phospho_pep in DephosphoPep_UniProtSeq_LUT[dephospho_pep]:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
525 if phospho_pep != phospho_pep:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
526 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
527 "phospho_pep:'%s' phospho_pep:'%s'"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
528 % (phospho_pep, phospho_pep)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
529 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
530 if phospho_pep not in PhosphoPep_UniProtSeq_LUT:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
531 PhosphoPep_UniProtSeq_LUT[phospho_pep] = phospho_pep
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
532 PhosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
533 (phospho_pep, DEPHOSPHOPEP)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
534 ] = dephospho_pep
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
535 r = list(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
536 zip(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
537 [s for s in UniProtSeqLUT[(sequence, UNIPROT_ID)]],
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
538 [s for s in UniProtSeqLUT[(sequence, GENE_NAME)]],
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
539 [
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
540 s
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
541 for s in UniProtSeqLUT[(sequence, DESCRIPTION)]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
542 ],
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
543 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
544 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
545 # Sort by `UniProt_ID`
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
546 # ref: https://stackoverflow.com/a/4174955/15509512
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
547 r = sorted(r, key=operator.itemgetter(0))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
548 # Get one tuple for each `phospho_pep`
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
549 # in DephosphoPep_UniProtSeq_LUT[dephospho_pep]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
550 for (upid, gn, desc) in r:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
551 # Append pseudo-tuple per UniProt_ID but only when it is not present
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
552 if (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
553 upid
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
554 not in DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
555 (dephospho_pep, UNIPROT_ID)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
556 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
557 ):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
558 DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
559 (dephospho_pep, UNIPROT_ID)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
560 ].append(upid)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
561 DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
562 (dephospho_pep, DESCRIPTION)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
563 ].append(desc)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
564 DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
565 (dephospho_pep, GENE_NAME)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
566 ].append(gn)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
567
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
568 # Close SwissProt SQLite database; clean up local variables
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
569 conn.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
570 # wipe local variables
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
571 phospho_pep = dephospho_pep = sequence = 0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
572 upid = gn = desc = r = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
573
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
574 # ----------- Get SwissProt data from SQLite database (finish) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
575
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
576 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
577 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
578 "%0.6f finished reading and decoding '%s' [0.4]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
579 % (end_time - start_time, upstream_map_filename_tab),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
580 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
581 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
582
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
583 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
584 "{:>10} unique upstream phosphopeptides tested".format(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
585 str(len(upstream_map_p_peptide_list))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
586 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
587 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
588
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
589 # Read in Upstream tabular file
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
590 # We are discarding the intensity data; so read it as text
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
591 upstream_data = pandas.read_table(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
592 upstream_map_filename_tab, dtype="str", index_col=0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
593 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
594
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
595 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
596 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
597 "%0.6f read Upstream Map from file [1g_1]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
598 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
599 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
600 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
601
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
602 upstream_data.index = upstream_map_p_peptide_list
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
603
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
604 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
605 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
606 "%0.6f added index to Upstream Map [1g_2]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
607 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
608 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
609 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
610
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
611 # trim upstream_data to include only the upstream map columns
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
612 old_cols = upstream_data.columns.tolist()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
613 i = 0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
614 first_intensity = -1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
615 last_intensity = -1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
616 intensity_re = re.compile("Intensity.*")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
617 for col_name in old_cols:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
618 m = intensity_re.match(col_name)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
619 if m:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
620 last_intensity = i
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
621 if first_intensity == -1:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
622 first_intensity = i
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
623 i += 1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
624 # print('last intensity = %d' % last_intensity)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
625 col_PKCalpha = last_intensity + 2
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
626
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
627 data_in_cols = [old_cols[0]] + old_cols[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
628 first_intensity: last_intensity + 1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
629 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
630
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
631 if upstream_data.empty:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
632 print("upstream_data is empty")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
633 exit(0)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
634
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
635 data_in = upstream_data.copy(deep=True)[data_in_cols]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
636
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
637 # Convert floating-point integers to int64 integers
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
638 # ref: https://stackoverflow.com/a/68497603/15509512
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
639 data_in[list(data_in.columns[1:])] = (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
640 data_in[list(data_in.columns[1:])]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
641 .astype("float64")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
642 .apply(np.int64)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
643 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
644
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
645 # create another phosphopeptide column that will be used to join later;
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
646 # MAY need to change depending on Phosphopeptide column position
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
647 # data_in[PHOSPHOPEPTIDE_MATCH] = data_in[data_in.columns.tolist()[0]]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
648 data_in[PHOSPHOPEPTIDE_MATCH] = data_in.index
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
649
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
650 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
651 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
652 "%0.6f set data_in[PHOSPHOPEPTIDE_MATCH] [A]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
653 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
654 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
655 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
656
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
657 # Produce a dictionary of metadata for a single phosphopeptide.
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
658 # This is a replacement of `UniProtInfo_subdict` in the original code.
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
659 def pseq_to_subdict(phospho_pep):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
660 # Strip "p" from phosphopeptide sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
661 dephospho_pep = re_phos.sub("", phospho_pep)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
662
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
663 # Determine number of phosphoresidues in phosphopeptide
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
664 numps = len(phospho_pep) - len(dephospho_pep)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
665
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
666 # Determine location(s) of phosphoresidue(s) in phosphopeptide
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
667 # (used later for Phosphoresidue, Sequence7, and Sequence10)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
668 ploc = [] # list of p locations
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
669 i = 0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
670 p = phospho_pep
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
671 while i < numps:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
672 ploc.append(p.find("p"))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
673 p = p[: p.find("p")] + p[p.find("p") + 1:]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
674 i += 1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
675
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
676 # Establish nested dictionary
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
677 result = {}
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
678 result[SEQUENCE] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
679 result[UNIPROT_ID] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
680 result[DESCRIPTION] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
681 result[GENE_NAME] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
682 result[PHOSPHORESIDUE] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
683 result[SEQUENCE7] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
684 result[SEQUENCE10] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
685
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
686 # Add stripped sequence to dictionary
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
687 result[SEQUENCE].append(dephospho_pep)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
688
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
689 # Locate phospho_pep in PhosphoPep_UniProtSeq_LUT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
690 # Caller may elect to:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
691 # try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
692 # ...
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
693 # except PreconditionError as pe:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
694 # print("'{expression}': {message}".format(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
695 # expression = pe.expression,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
696 # message = pe.message))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
697 # )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
698 # )
6
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
699 if phospho_pep not in PhosphoPep_UniProtSeq_LUT:
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
700 raise PreconditionError(
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
701 phospho_pep,
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
702 "no matching phosphopeptide found in PhosphoPep_UniProtSeq_LUT",
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
703 )
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
704 if dephospho_pep not in DephosphoPep_UniProtSeq_LUT:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
705 raise PreconditionError(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
706 dephospho_pep,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
707 "dephosphorylated phosphopeptide not found in DephosphoPep_UniProtSeq_LUT",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
708 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
709 if (
16
d9906288cc6a "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 7127d1511d45a8b22d4f50eee6db24d844371227"
eschen42
parents: 6
diff changeset
710 dephospho_pep != PhosphoPep_UniProtSeq_LUT[(phospho_pep, DEPHOSPHOPEP)]
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
711 ):
16
d9906288cc6a "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 7127d1511d45a8b22d4f50eee6db24d844371227"
eschen42
parents: 6
diff changeset
712 my_err_msg = "dephosphorylated phosphopeptide does not match "
d9906288cc6a "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 7127d1511d45a8b22d4f50eee6db24d844371227"
eschen42
parents: 6
diff changeset
713 my_err_msg += "PhosphoPep_UniProtSeq_LUT[(phospho_pep,DEPHOSPHOPEP)] = "
d9906288cc6a "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 7127d1511d45a8b22d4f50eee6db24d844371227"
eschen42
parents: 6
diff changeset
714 my_err_msg += PhosphoPep_UniProtSeq_LUT[(phospho_pep, DEPHOSPHOPEP)]
d9906288cc6a "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 7127d1511d45a8b22d4f50eee6db24d844371227"
eschen42
parents: 6
diff changeset
715 raise PreconditionError(dephospho_pep, my_err_msg)
d9906288cc6a "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 7127d1511d45a8b22d4f50eee6db24d844371227"
eschen42
parents: 6
diff changeset
716
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
717 result[SEQUENCE] = [dephospho_pep]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
718 result[UNIPROT_ID] = DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
719 (dephospho_pep, UNIPROT_ID)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
720 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
721 result[DESCRIPTION] = DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
722 (dephospho_pep, DESCRIPTION)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
723 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
724 result[GENE_NAME] = DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
725 (dephospho_pep, GENE_NAME)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
726 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
727 if (dephospho_pep, SEQUENCE) not in DephosphoPep_UniProtSeq_LUT:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
728 raise PreconditionError(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
729 dephospho_pep,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
730 "no matching phosphopeptide found in DephosphoPep_UniProtSeq_LUT",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
731 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
732 UniProtSeqList = DephosphoPep_UniProtSeq_LUT[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
733 (dephospho_pep, SEQUENCE)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
734 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
735 if len(UniProtSeqList) < 1:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
736 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
737 "Skipping DephosphoPep_UniProtSeq_LUT[('%s',SEQUENCE)] because value has zero length"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
738 % dephospho_pep
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
739 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
740 # raise PreconditionError(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
741 # "DephosphoPep_UniProtSeq_LUT[('" + dephospho_pep + ",SEQUENCE)",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
742 # 'value has zero length'
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
743 # )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
744 for UniProtSeq in UniProtSeqList:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
745 i = 0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
746 phosphoresidues = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
747 seq7s_set = set()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
748 seq7s = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
749 seq10s_set = set()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
750 seq10s = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
751 while i < len(ploc):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
752 start = UniProtSeq.find(dephospho_pep)
6
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
753 # handle case where no sequence was found for dep-pep
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
754 if start < 0:
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
755 i += 1
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
756 continue
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
757 psite = (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
758 start + ploc[i]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
759 ) # location of phosphoresidue on protein sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
760
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
761 # add Phosphoresidue
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
762 phosphosite = "p" + str(UniProtSeq)[psite] + str(psite + 1)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
763 phosphoresidues.append(phosphosite)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
764
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
765 # Add Sequence7
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
766 if psite < 7: # phospho_pep at N terminus
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
767 seq7 = str(UniProtSeq)[: psite + 8]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
768 if seq7[psite] == "S": # if phosphosresidue is serine
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
769 pres = "s"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
770 elif (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
771 seq7[psite] == "T"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
772 ): # if phosphosresidue is threonine
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
773 pres = "t"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
774 elif (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
775 seq7[psite] == "Y"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
776 ): # if phosphoresidue is tyrosine
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
777 pres = "y"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
778 else: # if not pSTY
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
779 pres = "?"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
780 seq7 = (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
781 seq7[:psite] + pres + seq7[psite + 1: psite + 8]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
782 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
783 while (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
784 len(seq7) < 15
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
785 ): # add appropriate number of "_" to the front
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
786 seq7 = "_" + seq7
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
787 elif (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
788 len(UniProtSeq) - psite < 8
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
789 ): # phospho_pep at C terminus
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
790 seq7 = str(UniProtSeq)[psite - 7:]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
791 if seq7[7] == "S":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
792 pres = "s"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
793 elif seq7[7] == "T":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
794 pres = "t"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
795 elif seq7[7] == "Y":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
796 pres = "y"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
797 else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
798 pres = "?"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
799 seq7 = seq7[:7] + pres + seq7[8:]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
800 while (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
801 len(seq7) < 15
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
802 ): # add appropriate number of "_" to the back
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
803 seq7 = seq7 + "_"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
804 else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
805 seq7 = str(UniProtSeq)[psite - 7: psite + 8]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
806 pres = "" # phosphoresidue
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
807 if seq7[7] == "S": # if phosphosresidue is serine
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
808 pres = "s"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
809 elif seq7[7] == "T": # if phosphosresidue is threonine
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
810 pres = "t"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
811 elif seq7[7] == "Y": # if phosphoresidue is tyrosine
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
812 pres = "y"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
813 else: # if not pSTY
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
814 pres = "?"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
815 seq7 = seq7[:7] + pres + seq7[8:]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
816 if seq7 not in seq7s_set:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
817 seq7s.append(seq7)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
818 seq7s_set.add(seq7)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
819
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
820 # add Sequence10
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
821 if psite < 10: # phospho_pep at N terminus
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
822 seq10 = (
16
d9906288cc6a "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 7127d1511d45a8b22d4f50eee6db24d844371227"
eschen42
parents: 6
diff changeset
823 str(UniProtSeq)[:psite] + "p" + str(UniProtSeq)[psite: psite + 11]
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
824 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
825 elif (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
826 len(UniProtSeq) - psite < 11
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
827 ): # phospho_pep at C terminus
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
828 seq10 = (
16
d9906288cc6a "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 7127d1511d45a8b22d4f50eee6db24d844371227"
eschen42
parents: 6
diff changeset
829 str(UniProtSeq)[psite - 10: psite] + "p" + str(UniProtSeq)[psite:]
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
830 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
831 else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
832 seq10 = str(UniProtSeq)[psite - 10: psite + 11]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
833 seq10 = seq10[:10] + "p" + seq10[10:]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
834 if seq10 not in seq10s_set:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
835 seq10s.append(seq10)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
836 seq10s_set.add(seq10)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
837
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
838 i += 1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
839
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
840 result[PHOSPHORESIDUE].append(phosphoresidues)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
841 result[SEQUENCE7].append(seq7s)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
842 # result[SEQUENCE10] is a list of lists of strings
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
843 result[SEQUENCE10].append(seq10s)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
844
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
845 r = list(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
846 zip(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
847 result[UNIPROT_ID],
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
848 result[GENE_NAME],
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
849 result[DESCRIPTION],
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
850 result[PHOSPHORESIDUE],
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
851 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
852 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
853 # Sort by `UniProt_ID`
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
854 # ref: https://stackoverflow.com//4174955/15509512
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
855 s = sorted(r, key=operator.itemgetter(0))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
856
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
857 result[UNIPROT_ID] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
858 result[GENE_NAME] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
859 result[DESCRIPTION] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
860 result[PHOSPHORESIDUE] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
861
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
862 for r in s:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
863 result[UNIPROT_ID].append(r[0])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
864 result[GENE_NAME].append(r[1])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
865 result[DESCRIPTION].append(r[2])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
866 result[PHOSPHORESIDUE].append(r[3])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
867
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
868 # convert lists to strings in the dictionary
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
869 for key, value in result.items():
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
870 if key not in [PHOSPHORESIDUE, SEQUENCE7, SEQUENCE10]:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
871 result[key] = "; ".join(map(str, value))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
872 elif key in [SEQUENCE10]:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
873 # result[SEQUENCE10] is a list of lists of strings
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
874 joined_value = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
875 joined_set = set()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
876 sep = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
877 for valL in value:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
878 # valL is a list of strings
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
879 for val in valL:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
880 # val is a string
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
881 if val not in joined_set:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
882 joined_set.add(val)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
883 joined_value += sep + val
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
884 sep = "; "
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
885 # joined_value is a string
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
886 result[key] = joined_value
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
887
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
888 newstring = "; ".join(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
889 [", ".join(prez) for prez in result[PHOSPHORESIDUE]]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
890 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
891 # #separate the isoforms in PHOSPHORESIDUE column with ";"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
892 # oldstring = result[PHOSPHORESIDUE]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
893 # oldlist = list(oldstring)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
894 # newstring = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
895 # i = 0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
896 # for e in oldlist:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
897 # if e == ";":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
898 # if numps > 1:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
899 # if i%numps:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
900 # newstring = newstring + ";"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
901 # else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
902 # newstring = newstring + ","
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
903 # else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
904 # newstring = newstring + ";"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
905 # i +=1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
906 # else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
907 # newstring = newstring + e
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
908 result[PHOSPHORESIDUE] = newstring
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
909
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
910 # separate sequence7's by |
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
911 oldstring = result[SEQUENCE7]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
912 oldlist = oldstring
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
913 newstring = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
914 for ol in oldlist:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
915 for e in ol:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
916 if e == ";":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
917 newstring = newstring + " |"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
918 elif len(newstring) > 0 and 1 > newstring.count(e):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
919 newstring = newstring + " | " + e
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
920 elif 1 > newstring.count(e):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
921 newstring = newstring + e
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
922 result[SEQUENCE7] = newstring
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
923
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
924 return [phospho_pep, result]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
925
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
926 # Construct list of [string, dictionary] lists
6
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
927 # where the dictionary provides the SwissProt metadata
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
928 # for a phosphopeptide
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
929 result_list = [
6
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
930 whine(pseq_to_subdict, psequence)
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
931 for psequence in data_in[PHOSPHOPEPTIDE_MATCH]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
932 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
933
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
934 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
935 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
936 "%0.6f added SwissProt annotations to phosphopeptides [B]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
937 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
938 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
939 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
940
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
941 # Construct dictionary from list of lists
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
942 # ref: https://www.8bitavenue.com/how-to-convert-list-of-lists-to-dictionary-in-python/
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
943 UniProt_Info = {
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
944 result[0]: result[1]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
945 for result in result_list
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
946 if result is not None
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
947 }
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
948
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
949 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
950 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
951 "%0.6f create dictionary mapping phosphopeptide to metadata dictionary [C]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
952 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
953 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
954 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
955
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
956 # cosmetic: add N_A to phosphopeptide rows with no hits
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
957 p_peptide_list = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
958 for key in UniProt_Info:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
959 p_peptide_list.append(key)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
960 for nestedKey in UniProt_Info[key]:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
961 if UniProt_Info[key][nestedKey] == "":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
962 UniProt_Info[key][nestedKey] = N_A
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
963
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
964 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
965 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
966 "%0.6f performed cosmetic clean-up [D]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
967 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
968 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
969
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
970 # convert UniProt_Info dictionary to dataframe
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
971 uniprot_df = pandas.DataFrame.transpose(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
972 pandas.DataFrame.from_dict(UniProt_Info)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
973 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
974
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
975 # reorder columns to match expected output file
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
976 uniprot_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
977 PHOSPHOPEPTIDE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
978 ] = uniprot_df.index # make index a column too
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
979
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
980 cols = uniprot_df.columns.tolist()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
981 # cols = [cols[-1]]+cols[4:6]+[cols[1]]+[cols[2]]+[cols[6]]+[cols[0]]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
982 # uniprot_df = uniprot_df[cols]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
983 uniprot_df = uniprot_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
984 [
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
985 PHOSPHOPEPTIDE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
986 SEQUENCE10,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
987 SEQUENCE7,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
988 GENE_NAME,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
989 PHOSPHORESIDUE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
990 UNIPROT_ID,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
991 DESCRIPTION,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
992 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
993 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
994
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
995 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
996 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
997 "%0.6f reordered columns to match expected output file [1]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
998 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
999 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1000 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1001
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1002 # concat to split then groupby to collapse
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1003 seq7_df = pandas.concat(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1004 [
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1005 pandas.Series(row[PHOSPHOPEPTIDE], row[SEQUENCE7].split(" | "))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1006 for _, row in uniprot_df.iterrows()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1007 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1008 ).reset_index()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1009 seq7_df.columns = [SEQUENCE7, PHOSPHOPEPTIDE]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1010
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1011 # --- -------------- begin read PSP_Regulatory_sites ---------------------------------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1012 # read in PhosphoSitePlus Regulatory Sites dataset
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1013 # ----------- Get PhosphoSitePlus Regulatory Sites data from SQLite database (start) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1014 conn = sql.connect(uniprot_sqlite)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1015 regsites_df = pandas.read_sql_query(PSP_REGSITE_SQL, conn)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1016 # Close SwissProt SQLite database
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1017 conn.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1018 # ... -------------- end read PSP_Regulatory_sites ------------------------------------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1019
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1020 # keep only the human entries in dataframe
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1021 if len(species) > 0:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1022 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1023 'Limit PhosphoSitesPlus records to species "' + species + '"'
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1024 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1025 regsites_df = regsites_df[regsites_df.ORGANISM == species]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1026
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1027 # merge the seq7 df with the regsites df based off of the sequence7
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1028 merge_df = seq7_df.merge(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1029 regsites_df,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1030 left_on=SEQUENCE7,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1031 right_on=SITE_PLUSMINUS_7AA_SQL,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1032 how="left",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1033 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1034
6
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
1035 # after merging df, select only the columns of interest;
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
1036 # note that PROTEIN is absent here
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1037 merge_df = merge_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1038 [
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1039 PHOSPHOPEPTIDE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1040 SEQUENCE7,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1041 ON_FUNCTION,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1042 ON_PROCESS,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1043 ON_PROT_INTERACT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1044 ON_OTHER_INTERACT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1045 ON_NOTES,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1046 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1047 ]
6
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
1048 # combine column values of interest
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
1049 # into one FUNCTION_PHOSPHORESIDUE column"
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1050 merge_df[FUNCTION_PHOSPHORESIDUE] = merge_df[ON_FUNCTION].str.cat(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1051 merge_df[ON_PROCESS], sep="; ", na_rep=""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1052 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1053 merge_df[FUNCTION_PHOSPHORESIDUE] = merge_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1054 FUNCTION_PHOSPHORESIDUE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1055 ].str.cat(merge_df[ON_PROT_INTERACT], sep="; ", na_rep="")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1056 merge_df[FUNCTION_PHOSPHORESIDUE] = merge_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1057 FUNCTION_PHOSPHORESIDUE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1058 ].str.cat(merge_df[ON_OTHER_INTERACT], sep="; ", na_rep="")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1059 merge_df[FUNCTION_PHOSPHORESIDUE] = merge_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1060 FUNCTION_PHOSPHORESIDUE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1061 ].str.cat(merge_df[ON_NOTES], sep="; ", na_rep="")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1062
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1063 # remove the columns that were combined
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1064 merge_df = merge_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1065 [PHOSPHOPEPTIDE, SEQUENCE7, FUNCTION_PHOSPHORESIDUE]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1066 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1067
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1068 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1069 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1070 "%0.6f merge regsite metadata [1a]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1071 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1072 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1073
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1074 # cosmetic changes to Function Phosphoresidue column
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1075 fp_series = pandas.Series(merge_df[FUNCTION_PHOSPHORESIDUE])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1076
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1077 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1078 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1079 "%0.6f more cosmetic changes [1b]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1080 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1081 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1082
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1083 i = 0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1084 while i < len(fp_series):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1085 # remove the extra ";" so that it looks more professional
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1086 if fp_series[i] == "; ; ; ; ": # remove ; from empty hits
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1087 fp_series[i] = ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1088 while fp_series[i].endswith("; "): # remove ; from the ends
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1089 fp_series[i] = fp_series[i][:-2]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1090 while fp_series[i].startswith("; "): # remove ; from the beginning
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1091 fp_series[i] = fp_series[i][2:]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1092 fp_series[i] = fp_series[i].replace("; ; ; ; ", "; ")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1093 fp_series[i] = fp_series[i].replace("; ; ; ", "; ")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1094 fp_series[i] = fp_series[i].replace("; ; ", "; ")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1095
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1096 # turn blanks into N_A to signify the info was searched for but cannot be found
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1097 if fp_series[i] == "":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1098 fp_series[i] = N_A
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1099
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1100 i += 1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1101 merge_df[FUNCTION_PHOSPHORESIDUE] = fp_series
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1102
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1103 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1104 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1105 "%0.6f cleaned up semicolons [1c]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1106 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1107 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1108
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1109 # merge uniprot df with merge df
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1110 uniprot_regsites_merged_df = uniprot_df.merge(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1111 merge_df,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1112 left_on=PHOSPHOPEPTIDE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1113 right_on=PHOSPHOPEPTIDE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1114 how="left",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1115 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1116
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1117 # collapse the merged df
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1118 uniprot_regsites_collapsed_df = pandas.DataFrame(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1119 uniprot_regsites_merged_df.groupby(PHOSPHOPEPTIDE)[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1120 FUNCTION_PHOSPHORESIDUE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1121 ].apply(lambda x: ppep_join(x))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1122 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1123 # .apply(lambda x: "%s" % ' | '.join(x)))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1124
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1125 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1126 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1127 "%0.6f collapsed pandas dataframe [1d]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1128 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1129 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1130
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1131 uniprot_regsites_collapsed_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1132 PHOSPHOPEPTIDE
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1133 ] = (
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1134 uniprot_regsites_collapsed_df.index
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1135 ) # add df index as its own column
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1136
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1137 # rename columns
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1138 uniprot_regsites_collapsed_df.columns = [
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1139 FUNCTION_PHOSPHORESIDUE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1140 "ppp",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1141 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1142
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1143 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1144 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1145 "%0.6f selected columns to be merged to uniprot_df [1e]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1146 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1147 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1148 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1149
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1150 # add columns based on Sequence7 matching site_+/-7_AA
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1151 uniprot_regsite_df = pandas.merge(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1152 left=uniprot_df,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1153 right=uniprot_regsites_collapsed_df,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1154 how="left",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1155 left_on=PHOSPHOPEPTIDE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1156 right_on="ppp",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1157 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1158
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1159 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1160 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1161 "%0.6f added columns based on Sequence7 matching site_+/-7_AA [1f]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1162 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1163 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1164 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1165
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1166 data_in.rename(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1167 {"Protein description": PHOSPHOPEPTIDE},
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1168 axis="columns",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1169 inplace=True,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1170 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1171
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1172 # data_in.sort_values(PHOSPHOPEPTIDE_MATCH, inplace=True, kind='mergesort')
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1173 res2 = sorted(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1174 data_in[PHOSPHOPEPTIDE_MATCH].tolist(), key=lambda s: s.casefold()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1175 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1176 data_in = data_in.loc[res2]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1177
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1178 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1179 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1180 "%0.6f sorting time [1f]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1181 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1182 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1183
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1184 cols = [old_cols[0]] + old_cols[col_PKCalpha - 1:]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1185 upstream_data = upstream_data[cols]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1186
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1187 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1188 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1189 "%0.6f refactored columns for Upstream Map [1g]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1190 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1191 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1192 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1193
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1194 # #rename upstream columns in new list
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1195 # new_cols = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1196 # for name in cols:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1197 # if "_NetworKIN" in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1198 # name = name.split("_")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1199 # if " motif" in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1200 # name = name.split(" motif")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1201 # if " sequence " in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1202 # name = name.split(" sequence")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1203 # if "_Phosida" in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1204 # name = name.split("_")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1205 # if "_PhosphoSite" in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1206 # name = name.split("_")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1207 # new_cols.append(name)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1208
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1209 # rename upstream columns in new list
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1210 def col_rename(name):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1211 if "_NetworKIN" in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1212 name = name.split("_")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1213 if " motif" in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1214 name = name.split(" motif")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1215 if " sequence " in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1216 name = name.split(" sequence")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1217 if "_Phosida" in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1218 name = name.split("_")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1219 if "_PhosphoSite" in name:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1220 name = name.split("_")[0]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1221 return name
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1222
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1223 new_cols = [col_rename(col) for col in cols]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1224 upstream_data.columns = new_cols
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1225
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1226 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1227 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1228 "%0.6f renamed columns for Upstream Map [1h_1]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1229 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1230 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1231 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1232
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1233 # Create upstream_data_cast as a copy of upstream_data
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1234 # but with first column substituted by the phosphopeptide sequence
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1235 upstream_data_cast = upstream_data.copy()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1236 new_cols_cast = new_cols
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1237 new_cols_cast[0] = "p_peptide"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1238 upstream_data_cast.columns = new_cols_cast
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1239 upstream_data_cast["p_peptide"] = upstream_data.index
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1240
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1241 # --- -------------- begin read upstream_data_melt ------------------------------------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1242 # ----------- Get melted kinase mapping data from SQLite database (start) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1243 conn = sql.connect(uniprot_sqlite)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1244 upstream_data_melt_df = pandas.read_sql_query(PPEP_MELT_SQL, conn)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1245 # Close SwissProt SQLite database
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1246 conn.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1247 upstream_data_melt = upstream_data_melt_df.copy()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1248 upstream_data_melt.columns = ["p_peptide", "characterization", "X"]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1249 upstream_data_melt["characterization"] = [
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1250 col_rename(s) for s in upstream_data_melt["characterization"]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1251 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1252
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1253 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1254 "%0.6f upstream_data_melt_df initially has %d rows"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1255 % (end_time - start_time, len(upstream_data_melt.axes[0])),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1256 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1257 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1258 # ref: https://stackoverflow.com/a/27360130/15509512
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1259 # e.g. df.drop(df[df.score < 50].index, inplace=True)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1260 upstream_data_melt.drop(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1261 upstream_data_melt[upstream_data_melt.X != "X"].index, inplace=True
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1262 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1263 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1264 "%0.6f upstream_data_melt_df pre-dedup has %d rows"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1265 % (end_time - start_time, len(upstream_data_melt.axes[0])),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1266 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1267 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1268 # ----------- Get melted kinase mapping data from SQLite database (finish) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1269 # ... -------------- end read upstream_data_melt --------------------------------------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1270
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1271 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1272 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1273 "%0.6f melted and minimized Upstream Map dataframe [1h_2]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1274 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1275 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1276 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1277 # ... end read upstream_data_melt
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1278
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1279 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1280 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1281 "%0.6f indexed melted Upstream Map [1h_2a]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1282 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1283 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1284 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1285
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1286 upstream_delta_melt_LoL = upstream_data_melt.values.tolist()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1287
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1288 melt_dict = {}
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1289 for key in upstream_map_p_peptide_list:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1290 melt_dict[key] = []
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1291
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1292 for el in upstream_delta_melt_LoL:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1293 (p_peptide, characterization, X) = tuple(el)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1294 if p_peptide in melt_dict:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1295 melt_dict[p_peptide].append(characterization)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1296 else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1297 exit(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1298 'Phosphopeptide %s not found in ppep_mapping_db: "phopsphopeptides" and "ppep_mapping_db" must both originate from the same run of mqppep_kinase_mapping'
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1299 % (p_peptide)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1300 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1301
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1302 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1303 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1304 "%0.6f appended peptide characterizations [1h_2b]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1305 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1306 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1307 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1308
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1309 # for key in upstream_map_p_peptide_list:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1310 # melt_dict[key] = ' | '.join(melt_dict[key])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1311
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1312 for key in upstream_map_p_peptide_list:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1313 melt_dict[key] = melt_join(melt_dict[key])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1314
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1315 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1316 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1317 "%0.6f concatenated multiple characterizations [1h_2c]"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1318 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1319 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1320 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1321
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1322 # map_dict is a dictionary of dictionaries
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1323 map_dict = {}
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1324 for key in upstream_map_p_peptide_list:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1325 map_dict[key] = {}
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1326 map_dict[key][PUTATIVE_UPSTREAM_DOMAINS] = melt_dict[key]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1327
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1328 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1329 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1330 "%0.6f instantiated map dictionary [2]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1331 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1332 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1333
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1334 # convert map_dict to dataframe
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1335 map_df = pandas.DataFrame.transpose(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1336 pandas.DataFrame.from_dict(map_dict)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1337 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1338 map_df["p-peptide"] = map_df.index # make index a column too
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1339 cols_map_df = map_df.columns.tolist()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1340 cols_map_df = [cols_map_df[1]] + [cols_map_df[0]]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1341 map_df = map_df[cols_map_df]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1342
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1343 # join map_df to uniprot_regsite_df
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1344 output_df = uniprot_regsite_df.merge(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1345 map_df, how="left", left_on=PHOSPHOPEPTIDE, right_on="p-peptide"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1346 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1347
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1348 output_df = output_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1349 [
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1350 PHOSPHOPEPTIDE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1351 SEQUENCE10,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1352 SEQUENCE7,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1353 GENE_NAME,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1354 PHOSPHORESIDUE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1355 UNIPROT_ID,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1356 DESCRIPTION,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1357 FUNCTION_PHOSPHORESIDUE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1358 PUTATIVE_UPSTREAM_DOMAINS,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1359 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1360 ]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1361
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1362 # cols_output_prelim = output_df.columns.tolist()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1363 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1364 # print("cols_output_prelim")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1365 # print(cols_output_prelim)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1366 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1367 # cols_output = cols_output_prelim[:8]+[cols_output_prelim[9]]+[cols_output_prelim[10]]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1368 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1369 # print("cols_output with p-peptide")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1370 # print(cols_output)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1371 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1372 # cols_output = [col for col in cols_output if not col == "p-peptide"]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1373 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1374 # print("cols_output")
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1375 # print(cols_output)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1376 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1377 # output_df = output_df[cols_output]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1378
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1379 # join output_df back to quantitative columns in data_in df
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1380 quant_cols = data_in.columns.tolist()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1381 quant_cols = quant_cols[1:]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1382 quant_data = data_in[quant_cols]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1383
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1384 # ----------- Write merge/filter metadata to SQLite database (start) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1385 # Open SwissProt SQLite database
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1386 conn = sql.connect(output_sqlite)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1387 cur = conn.cursor()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1388
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1389 cur.executescript(MRGFLTR_DDL)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1390
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1391 cur.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1392 CITATION_INSERT_STMT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1393 ("mrgfltr_metadata_view", CITATION_INSERT_PSP),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1394 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1395 cur.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1396 CITATION_INSERT_STMT, ("mrgfltr_metadata", CITATION_INSERT_PSP)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1397 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1398 cur.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1399 CITATION_INSERT_STMT,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1400 ("mrgfltr_metadata_view", CITATION_INSERT_PSP_REF),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1401 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1402 cur.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1403 CITATION_INSERT_STMT, ("mrgfltr_metadata", CITATION_INSERT_PSP_REF)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1404 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1405
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1406 # Read ppep-to-sequence LUT
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1407 ppep_lut_df = pandas.read_sql_query(PPEP_ID_SQL, conn)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1408 # write only metadata for merged/filtered records to SQLite
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1409 mrgfltr_metadata_df = output_df.copy()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1410 # replace phosphopeptide seq with ppep.id
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1411 mrgfltr_metadata_df = ppep_lut_df.merge(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1412 mrgfltr_metadata_df,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1413 left_on="ppep_seq",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1414 right_on=PHOSPHOPEPTIDE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1415 how="inner",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1416 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1417 mrgfltr_metadata_df.drop(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1418 columns=[PHOSPHOPEPTIDE, "ppep_seq"], inplace=True
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1419 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1420 # rename columns
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1421 mrgfltr_metadata_df.columns = MRGFLTR_METADATA_COLUMNS
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1422 mrgfltr_metadata_df.to_sql(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1423 "mrgfltr_metadata",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1424 con=conn,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1425 if_exists="append",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1426 index=False,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1427 method="multi",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1428 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1429
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1430 # Close SwissProt SQLite database
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1431 conn.close()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1432 # ----------- Write merge/filter metadata to SQLite database (finish) -----------
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1433
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1434 output_df = output_df.merge(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1435 quant_data,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1436 how="right",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1437 left_on=PHOSPHOPEPTIDE,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1438 right_on=PHOSPHOPEPTIDE_MATCH,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1439 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1440 output_cols = output_df.columns.tolist()
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1441 output_cols = output_cols[:-1]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1442 output_df = output_df[output_cols]
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1443
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1444 # cosmetic changes to Upstream column
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1445 output_df[PUTATIVE_UPSTREAM_DOMAINS] = output_df[
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1446 PUTATIVE_UPSTREAM_DOMAINS
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1447 ].fillna(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1448 ""
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1449 ) # fill the NaN with "" for those Phosphopeptides that got a "WARNING: Failed match for " in the upstream mapping
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1450 us_series = pandas.Series(output_df[PUTATIVE_UPSTREAM_DOMAINS])
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1451 i = 0
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1452 while i < len(us_series):
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1453 # turn blanks into N_A to signify the info was searched for but cannot be found
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1454 if us_series[i] == "":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1455 us_series[i] = N_A
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1456 i += 1
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1457 output_df[PUTATIVE_UPSTREAM_DOMAINS] = us_series
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1458
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1459 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1460 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1461 "%0.6f establisheed output [3]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1462 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1463 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1464
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1465 (output_rows, output_cols) = output_df.shape
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1466
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1467 output_df = output_df.convert_dtypes(convert_integer=True)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1468
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1469 # Output onto Final CSV file
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1470 output_df.to_csv(output_filename_csv, index=False)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1471 output_df.to_csv(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1472 output_filename_tab, quoting=None, sep="\t", index=False
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1473 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1474
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1475 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1476 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1477 "%0.6f wrote output [4]" % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1478 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1479 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1480
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1481 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1482 "{:>10} phosphopeptides written to output".format(str(output_rows))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1483 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1484
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1485 end_time = time.process_time() # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1486 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1487 "%0.6f seconds of non-system CPU time were consumed"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1488 % (end_time - start_time,),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1489 file=sys.stderr,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1490 ) # timer
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1491
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1492 # Rev. 7/1/2016
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1493 # Rev. 7/3/2016 : fill NaN in Upstream column to replace to N/A's
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1494 # Rev. 7/3/2016: renamed Upstream column to PUTATIVE_UPSTREAM_DOMAINS
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1495 # Rev. 12/2/2021: Converted to Python from ipynb; use fast Aho-Corasick searching; \
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1496 # read from SwissProt SQLite database
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1497 # Rev. 12/9/2021: Transfer code to Galaxy tool wrapper
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1498
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1499 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1500 # copied from Excel Output Script.ipynb END #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1501 #
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1502
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1503 try:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1504 catch(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1505 mqpep_getswissprot,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1506 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1507 exit(0)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1508 except Exception as e:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1509 exit("Internal error running mqpep_getswissprot(): %s" % (e))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1510
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1511
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1512 if __name__ == "__main__":
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
1513 __main__()