annotate search_ppep.py @ 0:d9b68bedbc91 draft

planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
author galaxyp
date Mon, 11 Jul 2022 19:20:41 +0000
parents
children 2276e88d5a1f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
1 #!/usr/bin/env python
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
2 # Search and memoize phosphopeptides in Swiss-Prot SQLite table UniProtKB
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
3
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
4 import argparse
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
5 import os.path
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
6 import re
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
7 import sqlite3
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
8 import sys # import the sys module for exc_info
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
9 import time
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
10 import traceback # import the traceback module for format_exception
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
11 from codecs import getreader as cx_getreader
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
12
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
13 # For Aho-Corasick search for fixed set of substrings
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
14 # - add_word
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
15 # - make_automaton
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
16 # - iter
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
17 import ahocorasick
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
18
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
19
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
20 # ref: https://stackoverflow.com/a/8915613/15509512
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
21 # answers: "How to handle exceptions in a list comprehensions"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
22 # usage:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
23 # from math import log
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
24 # eggs = [1,3,0,3,2]
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
25 # print([x for x in [catch(log, egg) for egg in eggs] if x is not None])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
26 # producing:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
27 # for <built-in function log>
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
28 # with args (0,)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
29 # exception: math domain error
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
30 # [0.0, 1.0986122886681098, 1.0986122886681098, 0.6931471805599453]
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
31 def catch(func, *args, handle=lambda e: e, **kwargs):
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
32
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
33 try:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
34 return func(*args, **kwargs)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
35 except Exception as e:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
36 print("For %s" % str(func))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
37 print(" with args %s" % str(args))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
38 print(" caught exception: %s" % str(e))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
39 (ty, va, tb) = sys.exc_info()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
40 print(" stack trace: " + str(traceback.format_exception(ty, va, tb)))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
41 # exit(-1)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
42 return None # was handle(e)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
43
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
44
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
45 def __main__():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
46
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
47 DROP_TABLES_SQL = """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
48 DROP VIEW IF EXISTS ppep_gene_site_view;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
49 DROP VIEW IF EXISTS uniprot_view;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
50 DROP VIEW IF EXISTS uniprotkb_pep_ppep_view;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
51 DROP VIEW IF EXISTS ppep_intensity_view;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
52 DROP VIEW IF EXISTS ppep_metadata_view;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
53
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
54 DROP TABLE IF EXISTS sample;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
55 DROP TABLE IF EXISTS ppep;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
56 DROP TABLE IF EXISTS site_type;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
57 DROP TABLE IF EXISTS deppep_UniProtKB;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
58 DROP TABLE IF EXISTS deppep;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
59 DROP TABLE IF EXISTS ppep_gene_site;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
60 DROP TABLE IF EXISTS ppep_metadata;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
61 DROP TABLE IF EXISTS ppep_intensity;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
62 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
63
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
64 CREATE_TABLES_SQL = """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
65 CREATE TABLE deppep
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
66 ( id INTEGER PRIMARY KEY
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
67 , seq TEXT UNIQUE ON CONFLICT IGNORE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
68 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
69 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
70 CREATE TABLE deppep_UniProtKB
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
71 ( deppep_id INTEGER REFERENCES deppep(id) ON DELETE CASCADE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
72 , UniProtKB_id TEXT REFERENCES UniProtKB(id) ON DELETE CASCADE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
73 , pos_start INTEGER
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
74 , pos_end INTEGER
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
75 , PRIMARY KEY (deppep_id, UniProtKB_id, pos_start, pos_end)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
76 ON CONFLICT IGNORE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
77 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
78 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
79 CREATE TABLE ppep
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
80 ( id INTEGER PRIMARY KEY
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
81 , deppep_id INTEGER REFERENCES deppep(id) ON DELETE CASCADE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
82 , seq TEXT UNIQUE ON CONFLICT IGNORE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
83 , scrubbed TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
84 );
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
85 CREATE TABLE site_type
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
86 ( id INTEGER PRIMARY KEY
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
87 , type_name TEXT UNIQUE ON CONFLICT IGNORE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
88 );
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
89 CREATE INDEX idx_ppep_scrubbed on ppep(scrubbed)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
90 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
91 CREATE TABLE sample
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
92 ( id INTEGER PRIMARY KEY
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
93 , name TEXT UNIQUE ON CONFLICT IGNORE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
94 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
95 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
96 CREATE VIEW uniprot_view AS
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
97 SELECT DISTINCT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
98 Uniprot_ID
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
99 , Description
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
100 , Organism_Name
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
101 , Organism_ID
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
102 , Gene_Name
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
103 , PE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
104 , SV
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
105 , Sequence
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
106 , Description ||
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
107 CASE WHEN Organism_Name = 'N/A'
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
108 THEN ''
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
109 ELSE ' OS='|| Organism_Name
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
110 END ||
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
111 CASE WHEN Organism_ID = -1
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
112 THEN ''
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
113 ELSE ' OX='|| Organism_ID
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
114 END ||
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
115 CASE WHEN Gene_Name = 'N/A'
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
116 THEN ''
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
117 ELSE ' GN='|| Gene_Name
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
118 END ||
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
119 CASE WHEN PE = 'N/A'
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
120 THEN ''
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
121 ELSE ' PE='|| PE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
122 END ||
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
123 CASE WHEN SV = 'N/A'
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
124 THEN ''
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
125 ELSE ' SV='|| SV
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
126 END AS long_description
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
127 , Database
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
128 FROM UniProtKB
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
129 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
130 CREATE VIEW uniprotkb_pep_ppep_view AS
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
131 SELECT deppep_UniProtKB.UniprotKB_ID AS accession
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
132 , deppep_UniProtKB.pos_start AS pos_start
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
133 , deppep_UniProtKB.pos_end AS pos_end
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
134 , deppep.seq AS peptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
135 , ppep.seq AS phosphopeptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
136 , ppep.scrubbed AS scrubbed
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
137 , uniprot_view.Sequence AS sequence
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
138 , uniprot_view.Description AS description
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
139 , uniprot_view.long_description AS long_description
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
140 , ppep.id AS ppep_id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
141 FROM ppep, deppep, deppep_UniProtKB, uniprot_view
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
142 WHERE deppep.id = ppep.deppep_id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
143 AND deppep.id = deppep_UniProtKB.deppep_id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
144 AND deppep_UniProtKB.UniprotKB_ID = uniprot_view.Uniprot_ID
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
145 ORDER BY UniprotKB_ID, deppep.seq, ppep.seq
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
146 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
147 CREATE TABLE ppep_gene_site
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
148 ( ppep_id INTEGER REFERENCES ppep(id)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
149 , gene_names TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
150 , site_type_id INTEGER REFERENCES site_type(id)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
151 , kinase_map TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
152 , PRIMARY KEY (ppep_id, kinase_map) ON CONFLICT IGNORE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
153 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
154 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
155 CREATE VIEW ppep_gene_site_view AS
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
156 SELECT DISTINCT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
157 ppep.seq AS phospho_peptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
158 , ppep_id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
159 , gene_names
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
160 , type_name
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
161 , kinase_map
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
162 FROM
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
163 ppep, ppep_gene_site, site_type
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
164 WHERE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
165 ppep_gene_site.ppep_id = ppep.id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
166 AND
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
167 ppep_gene_site.site_type_id = site_type.id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
168 ORDER BY
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
169 ppep.seq
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
170 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
171 CREATE TABLE ppep_metadata
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
172 ( ppep_id INTEGER REFERENCES ppep(id)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
173 , protein_description TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
174 , gene_name TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
175 , FASTA_name TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
176 , phospho_sites TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
177 , motifs_unique TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
178 , accessions TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
179 , motifs_all_members TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
180 , domain TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
181 , ON_FUNCTION TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
182 , ON_PROCESS TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
183 , ON_PROT_INTERACT TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
184 , ON_OTHER_INTERACT TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
185 , notes TEXT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
186 , PRIMARY KEY (ppep_id) ON CONFLICT IGNORE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
187 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
188 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
189 CREATE VIEW ppep_metadata_view AS
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
190 SELECT DISTINCT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
191 ppep.seq AS phospho_peptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
192 , protein_description
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
193 , gene_name
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
194 , FASTA_name
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
195 , phospho_sites
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
196 , motifs_unique
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
197 , accessions
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
198 , motifs_all_members
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
199 , domain
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
200 , ON_FUNCTION
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
201 , ON_PROCESS
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
202 , ON_PROT_INTERACT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
203 , ON_OTHER_INTERACT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
204 , notes
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
205 FROM
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
206 ppep, ppep_metadata
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
207 WHERE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
208 ppep_metadata.ppep_id = ppep.id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
209 ORDER BY
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
210 ppep.seq
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
211 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
212 CREATE TABLE ppep_intensity
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
213 ( ppep_id INTEGER REFERENCES ppep(id)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
214 , sample_id INTEGER
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
215 , intensity INTEGER
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
216 , PRIMARY KEY (ppep_id, sample_id) ON CONFLICT IGNORE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
217 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
218 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
219 CREATE VIEW ppep_intensity_view AS
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
220 SELECT DISTINCT
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
221 ppep.seq AS phospho_peptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
222 , sample.name AS sample
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
223 , intensity
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
224 FROM
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
225 ppep, sample, ppep_intensity
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
226 WHERE
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
227 ppep_intensity.sample_id = sample.id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
228 AND
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
229 ppep_intensity.ppep_id = ppep.id
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
230 ;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
231 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
232
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
233 UNIPROT_SEQ_AND_ID_SQL = """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
234 select Sequence, Uniprot_ID
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
235 from UniProtKB
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
236 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
237
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
238 # Parse Command Line
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
239 parser = argparse.ArgumentParser(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
240 description="Phopsphoproteomic Enrichment phosphopeptide SwissProt search (in place in SQLite DB)."
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
241 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
242
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
243 # inputs:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
244 # Phosphopeptide data for experimental results, including the intensities
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
245 # and the mapping to kinase domains, in tabular format.
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
246 parser.add_argument(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
247 "--phosphopeptides",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
248 "-p",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
249 nargs=1,
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
250 required=True,
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
251 dest="phosphopeptides",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
252 help="Phosphopeptide data for experimental results, generated by the Phopsphoproteomic Enrichment Localization Filter tool",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
253 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
254 parser.add_argument(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
255 "--uniprotkb",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
256 "-u",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
257 nargs=1,
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
258 required=True,
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
259 dest="uniprotkb",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
260 help="UniProtKB/Swiss-Prot data, converted from FASTA format by the Phopsphoproteomic Enrichment Kinase Mapping tool",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
261 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
262 parser.add_argument(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
263 "--schema",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
264 action="store_true",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
265 dest="db_schema",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
266 help="show updated database schema",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
267 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
268 parser.add_argument(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
269 "--warn-duplicates",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
270 action="store_true",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
271 dest="warn_duplicates",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
272 help="show warnings for duplicated sequences",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
273 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
274 parser.add_argument(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
275 "--verbose",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
276 action="store_true",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
277 dest="verbose",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
278 help="show somewhat verbose program tracing",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
279 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
280 # "Make it so!" (parse the arguments)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
281 options = parser.parse_args()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
282 if options.verbose:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
283 print("options: " + str(options) + "\n")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
284
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
285 # path to phosphopeptide (e.g., "outputfile_STEP2.txt") input tabular file
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
286 if options.phosphopeptides is None:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
287 exit('Argument "phosphopeptides" is required but not supplied')
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
288 try:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
289 f_name = os.path.abspath(options.phosphopeptides[0])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
290 except Exception as e:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
291 exit("Error parsing phosphopeptides argument: %s" % (e))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
292
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
293 # path to SQLite input/output tabular file
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
294 if options.uniprotkb is None:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
295 exit('Argument "uniprotkb" is required but not supplied')
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
296 try:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
297 db_name = os.path.abspath(options.uniprotkb[0])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
298 except Exception as e:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
299 exit("Error parsing uniprotkb argument: %s" % (e))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
300
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
301 # print("options.schema is %d" % options.db_schema)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
302
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
303 # db_name = "demo/test.sqlite"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
304 # f_name = "demo/test_input.txt"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
305
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
306 con = sqlite3.connect(db_name)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
307 cur = con.cursor()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
308 ker = con.cursor()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
309
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
310 cur.executescript(DROP_TABLES_SQL)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
311
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
312 # if options.db_schema:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
313 # print("\nAfter dropping tables/views that are to be created, schema is:")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
314 # cur.execute("SELECT * FROM sqlite_schema")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
315 # for row in cur.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
316 # if row[4] is not None:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
317 # print("%s;" % row[4])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
318
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
319 cur.executescript(CREATE_TABLES_SQL)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
320
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
321 if options.db_schema:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
322 print(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
323 "\nAfter creating tables/views that are to be created, schema is:"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
324 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
325 cur.execute("SELECT * FROM sqlite_schema")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
326 for row in cur.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
327 if row[4] is not None:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
328 print("%s;" % row[4])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
329
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
330 def generate_ppep(f):
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
331 # get keys from upstream tabular file using readline()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
332 # ref: https://stackoverflow.com/a/16713581/15509512
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
333 # answer to "Use codecs to read file with correct encoding"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
334 file1_encoded = open(f, "rb")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
335 file1 = cx_getreader("latin-1")(file1_encoded)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
336
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
337 count = 0
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
338 re_tab = re.compile("^[^\t]*")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
339 re_quote = re.compile('"')
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
340 while True:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
341 count += 1
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
342 # Get next line from file
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
343 line = file1.readline()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
344 # if line is empty
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
345 # end of file is reached
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
346 if not line:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
347 break
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
348 if count > 1:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
349 m = re_tab.match(line)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
350 m = re_quote.sub("", m[0])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
351 yield m
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
352 file1.close()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
353 file1_encoded.close()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
354
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
355 # Build an Aho-Corasick automaton from a trie
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
356 # - ref:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
357 # - https://pypi.org/project/pyahocorasick/
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
358 # - https://en.wikipedia.org/wiki/Aho%E2%80%93Corasick_algorithm
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
359 # - https://en.wikipedia.org/wiki/Trie
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
360 auto = ahocorasick.Automaton()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
361 re_phos = re.compile("p")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
362 # scrub out unsearchable characters per section
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
363 # "Match the p_peptides to the @sequences array:"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
364 # of the original
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
365 # PhosphoPeptide Upstream Kinase Mapping.pl
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
366 # which originally read
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
367 # $tmp_p_peptide =~ s/#//g;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
368 # $tmp_p_peptide =~ s/\d//g;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
369 # $tmp_p_peptide =~ s/\_//g;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
370 # $tmp_p_peptide =~ s/\.//g;
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
371 #
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
372 re_scrub = re.compile("0-9_.#")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
373 ppep_count = 0
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
374 for ppep in generate_ppep(f_name):
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
375 ppep_count += 1
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
376 add_to_trie = False
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
377 # print(ppep)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
378 scrubbed = re_scrub.sub("", ppep)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
379 deppep = re_phos.sub("", scrubbed)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
380 if options.verbose:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
381 print("deppep: %s; scrubbed: %s" % (deppep, scrubbed))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
382 # print(deppep)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
383 cur.execute("SELECT id FROM deppep WHERE seq = (?)", (deppep,))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
384 if cur.fetchone() is None:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
385 add_to_trie = True
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
386 cur.execute("INSERT INTO deppep(seq) VALUES (?)", (deppep,))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
387 cur.execute("SELECT id FROM deppep WHERE seq = (?)", (deppep,))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
388 deppep_id = cur.fetchone()[0]
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
389 if add_to_trie:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
390 # print((deppep_id, deppep))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
391 # Build the trie
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
392 auto.add_word(deppep, (deppep_id, deppep))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
393 cur.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
394 "INSERT INTO ppep(seq, scrubbed, deppep_id) VALUES (?,?,?)",
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
395 (ppep, scrubbed, deppep_id),
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
396 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
397 # def generate_deppep():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
398 # cur.execute("SELECT seq FROM deppep")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
399 # for row in cur.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
400 # yield row[0]
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
401 cur.execute("SELECT count(*) FROM (SELECT seq FROM deppep GROUP BY seq)")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
402 for row in cur.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
403 deppep_count = row[0]
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
404
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
405 cur.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
406 "SELECT count(*) FROM (SELECT Sequence FROM UniProtKB GROUP BY Sequence)"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
407 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
408 for row in cur.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
409 sequence_count = row[0]
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
410
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
411 print("%d phosphopeptides were read from input" % ppep_count)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
412 print(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
413 "%d corresponding dephosphopeptides are represented in input"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
414 % deppep_count
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
415 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
416 # Look for cases where both Gene_Name and Sequence are identical
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
417 cur.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
418 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
419 SELECT Uniprot_ID, Gene_Name, Sequence
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
420 FROM UniProtKB
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
421 WHERE Sequence IN (
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
422 SELECT Sequence
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
423 FROM UniProtKB
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
424 GROUP BY Sequence, Gene_Name
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
425 HAVING count(*) > 1
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
426 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
427 ORDER BY Sequence
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
428 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
429 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
430 duplicate_count = 0
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
431 old_seq = ""
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
432 for row in cur.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
433 if duplicate_count == 0:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
434 print(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
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)."
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
436 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
437 if row[2] != old_seq:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
438 old_seq = row[2]
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
439 duplicate_count += 1
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
440 if options.warn_duplicates:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
441 print("\n%s\t%s\t%s" % row)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
442 else:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
443 if options.warn_duplicates:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
444 print("%s\t%s" % (row[0], row[1]))
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
445 if duplicate_count > 0:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
446 print(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
447 "\n%d sequences have duplicated accession IDs\n" % duplicate_count
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
448 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
449
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
450 print("%s accession sequences will be searched\n" % sequence_count)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
451
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
452 # print(auto.dump())
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
453
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
454 # Convert the trie to an automaton (a finite-state machine)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
455 auto.make_automaton()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
456
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
457 # Execute query for seqs and metadata without fetching the results yet
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
458 uniprot_seq_and_id = cur.execute(UNIPROT_SEQ_AND_ID_SQL)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
459 while 1:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
460 batch = uniprot_seq_and_id.fetchmany(size=50)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
461 if not batch:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
462 break
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
463 for Sequence, UniProtKB_id in batch:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
464 if Sequence is not None:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
465 for end_index, (insert_order, original_value) in auto.iter(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
466 Sequence
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
467 ):
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
468 ker.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
469 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
470 INSERT INTO deppep_UniProtKB
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
471 (deppep_id,UniProtKB_id,pos_start,pos_end)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
472 VALUES (?,?,?,?)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
473 """,
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
474 (
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
475 insert_order,
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
476 UniProtKB_id,
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
477 1 + end_index - len(original_value),
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
478 end_index,
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
479 ),
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
480 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
481 else:
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
482 raise ValueError(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
483 "UniProtKB_id %s, but Sequence is None: Check whether SwissProt file is missing sequence for this ID"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
484 % (UniProtKB_id,)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
485 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
486 ker.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
487 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
488 SELECT count(*) || ' accession-peptide-phosphopeptide combinations were found'
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
489 FROM uniprotkb_pep_ppep_view
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
490 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
491 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
492 for row in ker.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
493 print(row[0])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
494
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
495 ker.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
496 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
497 SELECT count(*) || ' accession matches were found', count(*) AS accession_count
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
498 FROM (
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
499 SELECT accession
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
500 FROM uniprotkb_pep_ppep_view
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
501 GROUP BY accession
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
502 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
503 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
504 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
505 for row in ker.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
506 print(row[0])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
507
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
508 ker.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
509 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
510 SELECT count(*) || ' peptide matches were found'
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
511 FROM (
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
512 SELECT peptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
513 FROM uniprotkb_pep_ppep_view
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
514 GROUP BY peptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
515 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
516 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
517 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
518 for row in ker.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
519 print(row[0])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
520
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
521 ker.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
522 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
523 SELECT count(*) || ' phosphopeptide matches were found', count(*) AS phosphopeptide_count
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
524 FROM (
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
525 SELECT phosphopeptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
526 FROM uniprotkb_pep_ppep_view
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
527 GROUP BY phosphopeptide
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
528 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
529 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
530 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
531 for row in ker.fetchall():
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
532 print(row[0])
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
533
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
534 # link peptides not found in sequence database to a dummy sequence-record
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
535 ker.execute(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
536 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
537 INSERT INTO deppep_UniProtKB(deppep_id,UniProtKB_id,pos_start,pos_end)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
538 SELECT id, 'No Uniprot_ID', 0, 0
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
539 FROM deppep
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
540 WHERE id NOT IN (SELECT deppep_id FROM deppep_UniProtKB)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
541 """
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
542 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
543
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
544 con.commit()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
545 ker.execute("vacuum")
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
546 con.close()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
547
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
548
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
549 if __name__ == "__main__":
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
550 wrap_start_time = time.perf_counter()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
551 __main__()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
552 wrap_stop_time = time.perf_counter()
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
553 # print(wrap_start_time)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
554 # print(wrap_stop_time)
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
555 print(
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
556 "\nThe matching process took %d milliseconds to run.\n"
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
557 % ((wrap_stop_time - wrap_start_time) * 1000),
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
558 )
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
559
d9b68bedbc91 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit 3a7b3609d6e514c9e8f980ecb684960c6b2252fe
galaxyp
parents:
diff changeset
560 # vim: sw=4 ts=4 et ai :