annotate search_ppep.py @ 22:43f1fd0ff86b draft

"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 29775869faba
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
1 #!/usr/bin/env python
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
2 # Search and memoize phosphopeptides in Swiss-Prot SQLite table UniProtKB
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
3
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
4 import argparse
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
5 import os.path
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
6 import re
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
7 import sqlite3
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
8 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
9 import time
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
10 import traceback # import the traceback module for format_exception
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
11 from codecs import getreader as cx_getreader
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
12
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
13 # For Aho-Corasick search for fixed set of substrings
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
14 # - add_word
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
15 # - make_automaton
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
16 # - iter
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
17 import ahocorasick
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
18
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
19
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
20 # ref: https://stackoverflow.com/a/8915613/15509512
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
21 # answers: "How to handle exceptions in a list comprehensions"
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
22 # usage:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
23 # from math import log
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
24 # eggs = [1,3,0,3,2]
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
25 # print([x for x in [catch(log, egg) for egg in eggs] if x is not None])
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
26 # producing:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
27 # for <built-in function log>
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
28 # with args (0,)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
29 # exception: math domain error
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
30 # [0.0, 1.0986122886681098, 1.0986122886681098, 0.6931471805599453]
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
31 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
32
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
33 try:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
34 return func(*args, **kwargs)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
35 except Exception as e:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
36 print("For %s" % str(func))
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
37 print(" with args %s" % str(args))
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
38 print(" caught exception: %s" % str(e))
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
39 (ty, va, tb) = sys.exc_info()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
40 print(" stack trace: " + str(traceback.format_exception(ty, va, tb)))
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
41 # exit(-1)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
42 return None # was handle(e)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
43
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
44
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
45 def __main__():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
46
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
47 DROP_TABLES_SQL = """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
48 DROP VIEW IF EXISTS ppep_gene_site_view;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
49 DROP VIEW IF EXISTS uniprot_view;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
50 DROP VIEW IF EXISTS uniprotkb_pep_ppep_view;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
51 DROP VIEW IF EXISTS ppep_intensity_view;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
52 DROP VIEW IF EXISTS ppep_metadata_view;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
53
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
54 DROP TABLE IF EXISTS sample;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
55 DROP TABLE IF EXISTS ppep;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
56 DROP TABLE IF EXISTS site_type;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
57 DROP TABLE IF EXISTS deppep_UniProtKB;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
58 DROP TABLE IF EXISTS deppep;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
59 DROP TABLE IF EXISTS ppep_gene_site;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
60 DROP TABLE IF EXISTS ppep_metadata;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
61 DROP TABLE IF EXISTS ppep_intensity;
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
62 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
63
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
64 CREATE_TABLES_SQL = """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
65 CREATE TABLE deppep
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
66 ( id INTEGER PRIMARY KEY
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
67 , seq TEXT UNIQUE ON CONFLICT IGNORE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
68 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
69 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
70 CREATE TABLE deppep_UniProtKB
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
71 ( deppep_id INTEGER REFERENCES deppep(id) ON DELETE CASCADE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
72 , UniProtKB_id TEXT REFERENCES UniProtKB(id) ON DELETE CASCADE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
73 , pos_start INTEGER
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
74 , pos_end INTEGER
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
75 , PRIMARY KEY (deppep_id, UniProtKB_id, pos_start, pos_end)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
76 ON CONFLICT IGNORE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
77 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
78 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
79 CREATE TABLE ppep
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
80 ( id INTEGER PRIMARY KEY
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
81 , deppep_id INTEGER REFERENCES deppep(id) ON DELETE CASCADE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
82 , seq TEXT UNIQUE ON CONFLICT IGNORE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
83 , scrubbed TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
84 );
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
85 CREATE TABLE site_type
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
86 ( id INTEGER PRIMARY KEY
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
87 , type_name TEXT UNIQUE ON CONFLICT IGNORE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
88 );
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
89 CREATE INDEX idx_ppep_scrubbed on ppep(scrubbed)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
90 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
91 CREATE TABLE sample
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
92 ( id INTEGER PRIMARY KEY
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
93 , name TEXT UNIQUE ON CONFLICT IGNORE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
94 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
95 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
96 CREATE VIEW uniprot_view AS
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
97 SELECT DISTINCT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
98 Uniprot_ID
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
99 , Description
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
100 , Organism_Name
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
101 , Organism_ID
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
102 , Gene_Name
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
103 , PE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
104 , SV
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
105 , Sequence
17
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
106 , Description ||
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
107 CASE WHEN Organism_Name = 'N/A'
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
108 THEN ''
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
109 ELSE ' OS='|| Organism_Name
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
110 END ||
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
111 CASE WHEN Organism_ID = -1
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
112 THEN ''
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
113 ELSE ' OX='|| Organism_ID
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
114 END ||
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
115 CASE WHEN Gene_Name = 'N/A'
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
116 THEN ''
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
117 ELSE ' GN='|| Gene_Name
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
118 END ||
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
119 CASE WHEN PE = 'N/A'
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
120 THEN ''
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
121 ELSE ' PE='|| PE
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
122 END ||
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
123 CASE WHEN SV = 'N/A'
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
124 THEN ''
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
125 ELSE ' SV='|| SV
ba5f14c2a4af "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit aea9ac5a98069c3c993bd7903eda732f8ae2109d"
eschen42
parents: 6
diff changeset
126 END AS long_description
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
127 , Database
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
128 FROM UniProtKB
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
129 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
130 CREATE VIEW uniprotkb_pep_ppep_view AS
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
131 SELECT deppep_UniProtKB.UniprotKB_ID AS accession
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
132 , deppep_UniProtKB.pos_start AS pos_start
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
133 , deppep_UniProtKB.pos_end AS pos_end
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
134 , deppep.seq AS peptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
135 , ppep.seq AS phosphopeptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
136 , ppep.scrubbed AS scrubbed
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
137 , uniprot_view.Sequence AS sequence
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
138 , uniprot_view.Description AS description
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
139 , uniprot_view.long_description AS long_description
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
140 , ppep.id AS ppep_id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
141 FROM ppep, deppep, deppep_UniProtKB, uniprot_view
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
142 WHERE deppep.id = ppep.deppep_id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
143 AND deppep.id = deppep_UniProtKB.deppep_id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
144 AND deppep_UniProtKB.UniprotKB_ID = uniprot_view.Uniprot_ID
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
145 ORDER BY UniprotKB_ID, deppep.seq, ppep.seq
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
146 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
147 CREATE TABLE ppep_gene_site
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
148 ( ppep_id INTEGER REFERENCES ppep(id)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
149 , gene_names TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
150 , site_type_id INTEGER REFERENCES site_type(id)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
151 , kinase_map TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
152 , PRIMARY KEY (ppep_id, kinase_map) ON CONFLICT IGNORE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
153 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
154 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
155 CREATE VIEW ppep_gene_site_view AS
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
156 SELECT DISTINCT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
157 ppep.seq AS phospho_peptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
158 , ppep_id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
159 , gene_names
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
160 , type_name
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
161 , kinase_map
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
162 FROM
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
163 ppep, ppep_gene_site, site_type
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
164 WHERE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
165 ppep_gene_site.ppep_id = ppep.id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
166 AND
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
167 ppep_gene_site.site_type_id = site_type.id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
168 ORDER BY
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
169 ppep.seq
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
170 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
171 CREATE TABLE ppep_metadata
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
172 ( ppep_id INTEGER REFERENCES ppep(id)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
173 , protein_description TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
174 , gene_name TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
175 , FASTA_name TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
176 , phospho_sites TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
177 , motifs_unique TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
178 , accessions TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
179 , motifs_all_members TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
180 , domain TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
181 , ON_FUNCTION TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
182 , ON_PROCESS TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
183 , ON_PROT_INTERACT TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
184 , ON_OTHER_INTERACT TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
185 , notes TEXT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
186 , PRIMARY KEY (ppep_id) ON CONFLICT IGNORE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
187 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
188 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
189 CREATE VIEW ppep_metadata_view AS
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
190 SELECT DISTINCT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
191 ppep.seq AS phospho_peptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
192 , protein_description
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
193 , gene_name
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
194 , FASTA_name
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
195 , phospho_sites
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
196 , motifs_unique
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
197 , accessions
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
198 , motifs_all_members
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
199 , domain
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
200 , ON_FUNCTION
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
201 , ON_PROCESS
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
202 , ON_PROT_INTERACT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
203 , ON_OTHER_INTERACT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
204 , notes
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
205 FROM
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
206 ppep, ppep_metadata
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
207 WHERE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
208 ppep_metadata.ppep_id = ppep.id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
209 ORDER BY
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
210 ppep.seq
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
211 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
212 CREATE TABLE ppep_intensity
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
213 ( ppep_id INTEGER REFERENCES ppep(id)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
214 , sample_id INTEGER
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
215 , intensity INTEGER
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
216 , PRIMARY KEY (ppep_id, sample_id) ON CONFLICT IGNORE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
217 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
218 ;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
219 CREATE VIEW ppep_intensity_view AS
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
220 SELECT DISTINCT
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
221 ppep.seq AS phospho_peptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
222 , sample.name AS sample
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
223 , intensity
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
224 FROM
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
225 ppep, sample, ppep_intensity
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
226 WHERE
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
227 ppep_intensity.sample_id = sample.id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
228 AND
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
229 ppep_intensity.ppep_id = ppep.id
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
230 ;
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
231 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
232
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
233 UNIPROT_SEQ_AND_ID_SQL = """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
234 select Sequence, Uniprot_ID
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
235 from UniProtKB
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
236 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
237
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
238 # Parse Command Line
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
239 parser = argparse.ArgumentParser(
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
240 description="Phopsphoproteomic Enrichment phosphopeptide SwissProt search (in place in SQLite DB)."
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
241 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
242
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
243 # inputs:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
244 # Phosphopeptide data for experimental results, including the intensities
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
245 # and the mapping to kinase domains, in tabular format.
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
246 parser.add_argument(
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
247 "--phosphopeptides",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
248 "-p",
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
249 nargs=1,
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
250 required=True,
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
251 dest="phosphopeptides",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
252 help="Phosphopeptide data for experimental results, generated by the Phopsphoproteomic Enrichment Localization Filter tool",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
253 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
254 parser.add_argument(
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
255 "--uniprotkb",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
256 "-u",
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
257 nargs=1,
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
258 required=True,
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
259 dest="uniprotkb",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
260 help="UniProtKB/Swiss-Prot data, converted from FASTA format by the Phopsphoproteomic Enrichment Kinase Mapping tool",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
261 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
262 parser.add_argument(
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
263 "--schema",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
264 action="store_true",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
265 dest="db_schema",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
266 help="show updated database schema",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
267 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
268 parser.add_argument(
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
269 "--warn-duplicates",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
270 action="store_true",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
271 dest="warn_duplicates",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
272 help="show warnings for duplicated sequences",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
273 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
274 parser.add_argument(
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
275 "--verbose",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
276 action="store_true",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
277 dest="verbose",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
278 help="show somewhat verbose program tracing",
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
279 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
280 # "Make it so!" (parse the arguments)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
281 options = parser.parse_args()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
282 if options.verbose:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
283 print("options: " + str(options) + "\n")
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
284
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
285 # path to phosphopeptide (e.g., "outputfile_STEP2.txt") input tabular file
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
286 if options.phosphopeptides is None:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
287 exit('Argument "phosphopeptides" is required but not supplied')
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
288 try:
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
289 f_name = os.path.abspath(options.phosphopeptides[0])
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
290 except Exception as e:
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
291 exit("Error parsing phosphopeptides argument: %s" % (e))
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
292
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
293 # path to SQLite input/output tabular file
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
294 if options.uniprotkb is None:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
295 exit('Argument "uniprotkb" is required but not supplied')
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
296 try:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
297 db_name = os.path.abspath(options.uniprotkb[0])
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
298 except Exception as e:
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
299 exit("Error parsing uniprotkb argument: %s" % (e))
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
300
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
301 # print("options.schema is %d" % options.db_schema)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
302
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
303 # db_name = "demo/test.sqlite"
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
304 # f_name = "demo/test_input.txt"
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
305
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
306 con = sqlite3.connect(db_name)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
307 cur = con.cursor()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
308 ker = con.cursor()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
309
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
310 cur.executescript(DROP_TABLES_SQL)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
311
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
312 # if options.db_schema:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
313 # print("\nAfter dropping tables/views that are to be created, schema is:")
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
314 # cur.execute("SELECT * FROM sqlite_schema")
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
315 # for row in cur.fetchall():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
316 # if row[4] is not None:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
317 # print("%s;" % row[4])
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
318
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
319 cur.executescript(CREATE_TABLES_SQL)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
320
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
321 if options.db_schema:
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
322 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
323 "\nAfter creating tables/views that are to be created, schema is:"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
324 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
325 cur.execute("SELECT * FROM sqlite_schema")
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
326 for row in cur.fetchall():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
327 if row[4] is not None:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
328 print("%s;" % row[4])
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
329
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
330 def generate_ppep(f):
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
331 # get keys from upstream tabular file using readline()
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
332 # ref: https://stackoverflow.com/a/16713581/15509512
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
333 # answer to "Use codecs to read file with correct encoding"
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
334 file1_encoded = open(f, "rb")
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
335 file1 = cx_getreader("latin-1")(file1_encoded)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
336
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
337 count = 0
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
338 re_tab = re.compile("^[^\t]*")
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
339 re_quote = re.compile('"')
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
340 while True:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
341 count += 1
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
342 # Get next line from file
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
343 line = file1.readline()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
344 # if line is empty
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
345 # end of file is reached
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
346 if not line:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
347 break
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
348 if count > 1:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
349 m = re_tab.match(line)
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
350 m = re_quote.sub("", m[0])
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
351 yield m
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
352 file1.close()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
353 file1_encoded.close()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
354
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
355 # Build an Aho-Corasick automaton from a trie
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
356 # - ref:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
357 # - https://pypi.org/project/pyahocorasick/
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
358 # - https://en.wikipedia.org/wiki/Aho%E2%80%93Corasick_algorithm
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
359 # - https://en.wikipedia.org/wiki/Trie
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
360 auto = ahocorasick.Automaton()
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
361 re_phos = re.compile("p")
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
362 # scrub out unsearchable characters per section
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
363 # "Match the p_peptides to the @sequences array:"
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
364 # of the original
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
365 # PhosphoPeptide Upstream Kinase Mapping.pl
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
366 # which originally read
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
367 # $tmp_p_peptide =~ s/#//g;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
368 # $tmp_p_peptide =~ s/\d//g;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
369 # $tmp_p_peptide =~ s/\_//g;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
370 # $tmp_p_peptide =~ s/\.//g;
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
371 #
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
372 re_scrub = re.compile("0-9_.#")
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
373 ppep_count = 0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
374 for ppep in generate_ppep(f_name):
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
375 ppep_count += 1
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
376 add_to_trie = False
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
377 # print(ppep)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
378 scrubbed = re_scrub.sub("", ppep)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
379 deppep = re_phos.sub("", scrubbed)
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
380 if options.verbose:
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
381 print("deppep: %s; scrubbed: %s" % (deppep, scrubbed))
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
382 # print(deppep)
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
383 cur.execute("SELECT id FROM deppep WHERE seq = (?)", (deppep,))
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
384 if cur.fetchone() is None:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
385 add_to_trie = True
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
386 cur.execute("INSERT INTO deppep(seq) VALUES (?)", (deppep,))
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
387 cur.execute("SELECT id FROM deppep WHERE seq = (?)", (deppep,))
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
388 deppep_id = cur.fetchone()[0]
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
389 if add_to_trie:
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
390 # print((deppep_id, deppep))
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
391 # Build the trie
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
392 auto.add_word(deppep, (deppep_id, deppep))
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
393 cur.execute(
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
394 "INSERT INTO ppep(seq, scrubbed, deppep_id) VALUES (?,?,?)",
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
395 (ppep, scrubbed, deppep_id),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
396 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
397 # def generate_deppep():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
398 # cur.execute("SELECT seq FROM deppep")
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
399 # for row in cur.fetchall():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
400 # yield row[0]
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
401 cur.execute("SELECT count(*) FROM (SELECT seq FROM deppep GROUP BY seq)")
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
402 for row in cur.fetchall():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
403 deppep_count = row[0]
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
404
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
405 cur.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
406 "SELECT count(*) FROM (SELECT Sequence FROM UniProtKB 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
407 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
408 for row in cur.fetchall():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
409 sequence_count = row[0]
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
410
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
411 print("%d phosphopeptides were read from input" % ppep_count)
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
412 print(
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
413 "%d corresponding dephosphopeptides are represented in input"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
414 % deppep_count
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
415 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
416 # Look for cases where both Gene_Name and Sequence are identical
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
417 cur.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
418 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
419 SELECT Uniprot_ID, Gene_Name, Sequence
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
420 FROM UniProtKB
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
421 WHERE Sequence IN (
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
422 SELECT Sequence
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
423 FROM UniProtKB
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
424 GROUP BY Sequence, Gene_Name
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
425 HAVING count(*) > 1
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
426 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
427 ORDER BY Sequence
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
428 """
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
429 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
430 duplicate_count = 0
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
431 old_seq = ""
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
432 for row in cur.fetchall():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
433 if duplicate_count == 0:
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
434 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
435 "\nEach of the following sequences is associated with several accession IDs (which are listed in the first column) but the same gene ID (which is listed in the second column)."
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
436 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
437 if row[2] != old_seq:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
438 old_seq = row[2]
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
439 duplicate_count += 1
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
440 if options.warn_duplicates:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
441 print("\n%s\t%s\t%s" % row)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
442 else:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
443 if options.warn_duplicates:
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
444 print("%s\t%s" % (row[0], row[1]))
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
445 if duplicate_count > 0:
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
446 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
447 "\n%d sequences have duplicated accession IDs\n" % duplicate_count
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
448 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
449
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
450 print("%s accession sequences will be searched\n" % sequence_count)
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
451
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
452 # print(auto.dump())
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
453
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
454 # Convert the trie to an automaton (a finite-state machine)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
455 auto.make_automaton()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
456
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
457 # Execute query for seqs and metadata without fetching the results yet
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
458 uniprot_seq_and_id = cur.execute(UNIPROT_SEQ_AND_ID_SQL)
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
459 while 1:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
460 batch = uniprot_seq_and_id.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
461 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
462 break
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
463 for Sequence, UniProtKB_id in batch:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
464 if Sequence 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
465 for end_index, (insert_order, original_value) in auto.iter(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
466 Sequence
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 ker.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
469 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
470 INSERT INTO deppep_UniProtKB
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
471 (deppep_id,UniProtKB_id,pos_start,pos_end)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
472 VALUES (?,?,?,?)
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
473 """,
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 insert_order,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
476 UniProtKB_id,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
477 1 + end_index - len(original_value),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
478 end_index,
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
479 ),
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 else:
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
482 raise ValueError(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
483 "UniProtKB_id %s, but Sequence is None: Check whether SwissProt file is missing sequence for this ID"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
484 % (UniProtKB_id,)
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
485 )
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
486 ker.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
487 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
488 SELECT count(*) || ' accession-peptide-phosphopeptide combinations were found'
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
489 FROM uniprotkb_pep_ppep_view
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
490 """
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
491 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
492 for row in ker.fetchall():
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
493 print(row[0])
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
494
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
495 ker.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
496 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
497 SELECT count(*) || ' accession matches were found', count(*) AS accession_count
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
498 FROM (
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
499 SELECT accession
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
500 FROM uniprotkb_pep_ppep_view
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
501 GROUP BY accession
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
502 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
503 """
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
504 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
505 for row in ker.fetchall():
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
506 print(row[0])
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
507
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
508 ker.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
509 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
510 SELECT count(*) || ' peptide matches were found'
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
511 FROM (
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
512 SELECT peptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
513 FROM uniprotkb_pep_ppep_view
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
514 GROUP BY peptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
515 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
516 """
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
517 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
518 for row in ker.fetchall():
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
519 print(row[0])
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
520
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
521 ker.execute(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
522 """
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
523 SELECT count(*) || ' phosphopeptide matches were found', count(*) AS phosphopeptide_count
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
524 FROM (
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
525 SELECT phosphopeptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
526 FROM uniprotkb_pep_ppep_view
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
527 GROUP BY phosphopeptide
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
528 )
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
529 """
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
530 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
531 for row in ker.fetchall():
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
532 print(row[0])
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
533
6
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
534 # link peptides not found in sequence database to a dummy sequence-record
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
535 ker.execute(
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
536 """
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
537 INSERT INTO deppep_UniProtKB(deppep_id,UniProtKB_id,pos_start,pos_end)
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
538 SELECT id, 'No Uniprot_ID', 0, 0
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
539 FROM deppep
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
540 WHERE id NOT IN (SELECT deppep_id FROM deppep_UniProtKB)
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
541 """
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
542 )
42daf70d4ed4 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 9dfb7e07a3673d7de4b0a1b7e6ce1b75a8a4f42b"
eschen42
parents: 5
diff changeset
543
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
544 con.commit()
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
545 ker.execute("vacuum")
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
546 con.close()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
547
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
548
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
549 if __name__ == "__main__":
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
550 wrap_start_time = time.perf_counter()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
551 __main__()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
552 wrap_stop_time = time.perf_counter()
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
553 # print(wrap_start_time)
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
554 # print(wrap_stop_time)
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
555 print(
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
556 "\nThe matching process took %d milliseconds to run.\n"
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
557 % ((wrap_stop_time - wrap_start_time) * 1000),
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
558 )
0
2c7e1b167736 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff changeset
559
5
b91809a18dbe "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 92e8ab6fc27a1f02583742715d644bc96418fbdf"
eschen42
parents: 0
diff changeset
560 # vim: sw=4 ts=4 et ai :