Mercurial > repos > eschen42 > mqppep_anova
annotate search_ppep.py @ 0:c1403d18c189 draft
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
author | eschen42 |
---|---|
date | Mon, 07 Mar 2022 19:05:01 +0000 |
parents | |
children | d4d531006735 |
rev | line source |
---|---|
0
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
1 #!/usr/bin/env python |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
2 # Search and memoize phosphopeptides in Swiss-Prot SQLite table UniProtKB |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
3 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
4 import argparse |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
5 import os.path |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
6 import sqlite3 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
7 import re |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
8 from codecs import getreader as cx_getreader |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
9 import time |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
10 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
11 # For Aho-Corasick search for fixed set of substrings |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
12 # - add_word |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
13 # - make_automaton |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
14 # - iter |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
15 import ahocorasick |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
16 # Support map over auto.iter(...) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
17 # - itemgetter |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
18 import operator |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
19 #import hashlib |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
20 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
21 # ref: https://stackoverflow.com/a/8915613/15509512 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
22 # answers: "How to handle exceptions in a list comprehensions" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
23 # usage: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
24 # from math import log |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
25 # eggs = [1,3,0,3,2] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
26 # print([x for x in [catch(log, egg) for egg in eggs] if x is not None]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
27 # producing: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
28 # for <built-in function log> |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
29 # with args (0,) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
30 # exception: math domain error |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
31 # [0.0, 1.0986122886681098, 1.0986122886681098, 0.6931471805599453] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
32 def catch(func, *args, handle=lambda e : e, **kwargs): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
33 try: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
34 return func(*args, **kwargs) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
35 except Exception as e: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
36 print("For %s" % str(func)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
37 print(" with args %s" % str(args)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
38 print(" caught exception: %s" % str(e)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
39 (ty, va, tb) = sys.exc_info() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
40 print(" stack trace: " + str(traceback.format_exception(ty, va, tb))) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
41 #exit(-1) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
42 return None # was handle(e) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
43 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
44 def __main__(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
45 ITEM_GETTER = operator.itemgetter(1) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
46 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
47 DROP_TABLES_SQL = ''' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
48 DROP VIEW IF EXISTS ppep_gene_site_view; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
49 DROP VIEW IF EXISTS uniprot_view; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
50 DROP VIEW IF EXISTS uniprotkb_pep_ppep_view; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
51 DROP VIEW IF EXISTS ppep_intensity_view; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
52 DROP VIEW IF EXISTS ppep_metadata_view; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
53 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
54 DROP TABLE IF EXISTS sample; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
55 DROP TABLE IF EXISTS ppep; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
56 DROP TABLE IF EXISTS site_type; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
57 DROP TABLE IF EXISTS deppep_UniProtKB; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
58 DROP TABLE IF EXISTS deppep; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
59 DROP TABLE IF EXISTS ppep_gene_site; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
60 DROP TABLE IF EXISTS ppep_metadata; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
61 DROP TABLE IF EXISTS ppep_intensity; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
62 ''' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
63 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
64 CREATE_TABLES_SQL = ''' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
65 CREATE TABLE deppep |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
66 ( id INTEGER PRIMARY KEY |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
67 , seq TEXT UNIQUE ON CONFLICT IGNORE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
68 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
69 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
70 CREATE TABLE deppep_UniProtKB |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
71 ( deppep_id INTEGER REFERENCES deppep(id) ON DELETE CASCADE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
72 , UniProtKB_id TEXT REFERENCES UniProtKB(id) ON DELETE CASCADE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
73 , pos_start INTEGER |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
74 , pos_end INTEGER |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
75 , PRIMARY KEY (deppep_id, UniProtKB_id, pos_start, pos_end) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
76 ON CONFLICT IGNORE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
77 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
78 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
79 CREATE TABLE ppep |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
80 ( id INTEGER PRIMARY KEY |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
81 , deppep_id INTEGER REFERENCES deppep(id) ON DELETE CASCADE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
82 , seq TEXT UNIQUE ON CONFLICT IGNORE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
83 , scrubbed TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
84 ); |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
85 CREATE TABLE site_type |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
86 ( id INTEGER PRIMARY KEY |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
87 , type_name TEXT UNIQUE ON CONFLICT IGNORE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
88 ); |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
89 CREATE INDEX idx_ppep_scrubbed on ppep(scrubbed) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
90 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
91 CREATE TABLE sample |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
92 ( id INTEGER PRIMARY KEY |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
93 , name TEXT UNIQUE ON CONFLICT IGNORE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
94 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
95 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
96 CREATE VIEW uniprot_view AS |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
97 SELECT DISTINCT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
98 Uniprot_ID |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
99 , Description |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
100 , Organism_Name |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
101 , Organism_ID |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
102 , Gene_Name |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
103 , PE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
104 , SV |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
105 , Sequence |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
106 , Description || ' OS=' || |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
107 Organism_Name || ' OX=' || Organism_ID || |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
108 CASE WHEN Gene_Name = 'N/A' THEN '' ELSE ' GN='|| Gene_Name END || |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
109 CASE WHEN PE = 'N/A' THEN '' ELSE ' PE='|| PE END || |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
110 CASE WHEN SV = 'N/A' THEN '' ELSE ' SV='|| SV END |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
111 AS long_description |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
112 , Database |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
113 FROM UniProtKB |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
114 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
115 CREATE VIEW uniprotkb_pep_ppep_view AS |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
116 SELECT deppep_UniProtKB.UniprotKB_ID AS accession |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
117 , deppep_UniProtKB.pos_start AS pos_start |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
118 , deppep_UniProtKB.pos_end AS pos_end |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
119 , deppep.seq AS peptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
120 , ppep.seq AS phosphopeptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
121 , ppep.scrubbed AS scrubbed |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
122 , uniprot_view.Sequence AS sequence |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
123 , uniprot_view.Description AS description |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
124 , uniprot_view.long_description AS long_description |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
125 , ppep.id AS ppep_id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
126 FROM ppep, deppep, deppep_UniProtKB, uniprot_view |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
127 WHERE deppep.id = ppep.deppep_id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
128 AND deppep.id = deppep_UniProtKB.deppep_id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
129 AND deppep_UniProtKB.UniprotKB_ID = uniprot_view.Uniprot_ID |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
130 ORDER BY UniprotKB_ID, deppep.seq, ppep.seq |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
131 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
132 CREATE TABLE ppep_gene_site |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
133 ( ppep_id INTEGER REFERENCES ppep(id) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
134 , gene_names TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
135 , site_type_id INTEGER REFERENCES site_type(id) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
136 , kinase_map TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
137 , PRIMARY KEY (ppep_id, kinase_map) ON CONFLICT IGNORE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
138 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
139 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
140 CREATE VIEW ppep_gene_site_view AS |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
141 SELECT DISTINCT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
142 ppep.seq AS phospho_peptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
143 , ppep_id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
144 , gene_names |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
145 , type_name |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
146 , kinase_map |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
147 FROM |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
148 ppep, ppep_gene_site, site_type |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
149 WHERE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
150 ppep_gene_site.ppep_id = ppep.id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
151 AND |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
152 ppep_gene_site.site_type_id = site_type.id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
153 ORDER BY |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
154 ppep.seq |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
155 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
156 CREATE TABLE ppep_metadata |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
157 ( ppep_id INTEGER REFERENCES ppep(id) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
158 , protein_description TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
159 , gene_name TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
160 , FASTA_name TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
161 , phospho_sites TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
162 , motifs_unique TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
163 , accessions TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
164 , motifs_all_members TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
165 , domain TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
166 , ON_FUNCTION TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
167 , ON_PROCESS TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
168 , ON_PROT_INTERACT TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
169 , ON_OTHER_INTERACT TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
170 , notes TEXT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
171 , PRIMARY KEY (ppep_id) ON CONFLICT IGNORE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
172 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
173 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
174 CREATE VIEW ppep_metadata_view AS |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
175 SELECT DISTINCT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
176 ppep.seq AS phospho_peptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
177 , protein_description |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
178 , gene_name |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
179 , FASTA_name |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
180 , phospho_sites |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
181 , motifs_unique |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
182 , accessions |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
183 , motifs_all_members |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
184 , domain |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
185 , ON_FUNCTION |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
186 , ON_PROCESS |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
187 , ON_PROT_INTERACT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
188 , ON_OTHER_INTERACT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
189 , notes |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
190 FROM |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
191 ppep, ppep_metadata |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
192 WHERE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
193 ppep_metadata.ppep_id = ppep.id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
194 ORDER BY |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
195 ppep.seq |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
196 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
197 CREATE TABLE ppep_intensity |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
198 ( ppep_id INTEGER REFERENCES ppep(id) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
199 , sample_id INTEGER |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
200 , intensity INTEGER |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
201 , PRIMARY KEY (ppep_id, sample_id) ON CONFLICT IGNORE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
202 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
203 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
204 CREATE VIEW ppep_intensity_view AS |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
205 SELECT DISTINCT |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
206 ppep.seq AS phospho_peptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
207 , sample.name AS sample |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
208 , intensity |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
209 FROM |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
210 ppep, sample, ppep_intensity |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
211 WHERE |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
212 ppep_intensity.sample_id = sample.id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
213 AND |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
214 ppep_intensity.ppep_id = ppep.id |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
215 ; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
216 ''' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
217 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
218 UNIPROT_SEQ_AND_ID_SQL = ''' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
219 select Sequence, Uniprot_ID |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
220 from UniProtKB |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
221 ''' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
222 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
223 # Parse Command Line |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
224 parser = argparse.ArgumentParser( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
225 description='Phopsphoproteomic Enrichment phosphopeptide SwissProt search (in place in SQLite DB).' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
226 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
227 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
228 # inputs: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
229 # Phosphopeptide data for experimental results, including the intensities |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
230 # and the mapping to kinase domains, in tabular format. |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
231 parser.add_argument( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
232 '--phosphopeptides', '-p', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
233 nargs=1, |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
234 required=True, |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
235 dest='phosphopeptides', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
236 help='Phosphopeptide data for experimental results, generated by the Phopsphoproteomic Enrichment Localization Filter tool' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
237 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
238 parser.add_argument( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
239 '--uniprotkb', '-u', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
240 nargs=1, |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
241 required=True, |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
242 dest='uniprotkb', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
243 help='UniProtKB/Swiss-Prot data, converted from FASTA format by the Phopsphoproteomic Enrichment Kinase Mapping tool' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
244 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
245 parser.add_argument( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
246 '--schema', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
247 action='store_true', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
248 dest='db_schema', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
249 help='show updated database schema' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
250 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
251 parser.add_argument( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
252 '--warn-duplicates', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
253 action='store_true', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
254 dest='warn_duplicates', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
255 help='show warnings for duplicated sequences' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
256 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
257 parser.add_argument( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
258 '--verbose', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
259 action='store_true', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
260 dest='verbose', |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
261 help='show somewhat verbose program tracing' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
262 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
263 # "Make it so!" (parse the arguments) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
264 options = parser.parse_args() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
265 if options.verbose: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
266 print("options: " + str(options) + "\n") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
267 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
268 # path to phosphopeptide (e.g., "outputfile_STEP2.txt") input tabular file |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
269 if options.phosphopeptides is None: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
270 exit('Argument "phosphopeptides" is required but not supplied') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
271 try: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
272 f_name = os.path.abspath(options.phosphopeptides[0]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
273 except Exception as e: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
274 exit('Error parsing phosphopeptides argument: %s' % (e)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
275 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
276 # path to SQLite input/output tabular file |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
277 if options.uniprotkb is None: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
278 exit('Argument "uniprotkb" is required but not supplied') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
279 try: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
280 db_name = os.path.abspath(options.uniprotkb[0]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
281 except Exception as e: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
282 exit('Error parsing uniprotkb argument: %s' % (e)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
283 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
284 # print("options.schema is %d" % options.db_schema) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
285 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
286 # db_name = "demo/test.sqlite" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
287 # f_name = "demo/test_input.txt" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
288 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
289 con = sqlite3.connect(db_name) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
290 cur = con.cursor() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
291 ker = con.cursor() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
292 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
293 cur.executescript(DROP_TABLES_SQL) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
294 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
295 # if options.db_schema: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
296 # print("\nAfter dropping tables/views that are to be created, schema is:") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
297 # cur.execute("SELECT * FROM sqlite_schema") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
298 # for row in cur.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
299 # if row[4] is not None: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
300 # print("%s;" % row[4]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
301 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
302 cur.executescript(CREATE_TABLES_SQL) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
303 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
304 if options.db_schema: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
305 print("\nAfter creating tables/views that are to be created, schema is:") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
306 cur.execute("SELECT * FROM sqlite_schema") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
307 for row in cur.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
308 if row[4] is not None: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
309 print("%s;" % row[4]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
310 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
311 def generate_ppep(f): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
312 #get keys from upstream tabular file using readline() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
313 # ref: https://stackoverflow.com/a/16713581/15509512 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
314 # answer to "Use codecs to read file with correct encoding" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
315 file1_encoded = open(f, 'rb') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
316 file1 = cx_getreader("latin-1")(file1_encoded) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
317 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
318 count = 0 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
319 re_tab = re.compile('^[^\t]*') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
320 re_quote = re.compile('"') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
321 while True: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
322 count += 1 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
323 # Get next line from file |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
324 line = file1.readline() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
325 # if line is empty |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
326 # end of file is reached |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
327 if not line: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
328 break |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
329 if count > 1: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
330 m = re_tab.match(line) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
331 m = re_quote.sub('',m[0]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
332 yield m |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
333 file1.close() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
334 file1_encoded.close() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
335 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
336 # Build an Aho-Corasick automaton from a trie |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
337 # - ref: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
338 # - https://pypi.org/project/pyahocorasick/ |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
339 # - https://en.wikipedia.org/wiki/Aho%E2%80%93Corasick_algorithm |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
340 # - https://en.wikipedia.org/wiki/Trie |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
341 auto = ahocorasick.Automaton() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
342 re_phos = re.compile('p') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
343 # scrub out unsearchable characters per section |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
344 # "Match the p_peptides to the @sequences array:" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
345 # of the original |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
346 # PhosphoPeptide Upstream Kinase Mapping.pl |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
347 # which originally read |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
348 # $tmp_p_peptide =~ s/#//g; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
349 # $tmp_p_peptide =~ s/\d//g; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
350 # $tmp_p_peptide =~ s/\_//g; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
351 # $tmp_p_peptide =~ s/\.//g; |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
352 # |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
353 re_scrub = re.compile('0-9_.#') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
354 ppep_count = 0 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
355 for ppep in generate_ppep(f_name): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
356 ppep_count += 1 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
357 add_to_trie = False |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
358 #print(ppep) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
359 scrubbed = re_scrub.sub('',ppep) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
360 deppep = re_phos.sub('',scrubbed) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
361 if options.verbose: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
362 print("deppep: %s; scrubbed: %s" % (deppep,scrubbed)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
363 #print(deppep) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
364 cur.execute("SELECT id FROM deppep WHERE seq = (?)", (deppep,)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
365 if cur.fetchone() is None: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
366 add_to_trie = True |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
367 cur.execute("INSERT INTO deppep(seq) VALUES (?)", (deppep,)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
368 cur.execute("SELECT id FROM deppep WHERE seq = (?)", (deppep,)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
369 deppep_id = cur.fetchone()[0] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
370 if add_to_trie: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
371 #print((deppep_id, deppep)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
372 # Build the trie |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
373 auto.add_word(deppep, (deppep_id, deppep)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
374 cur.execute( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
375 "INSERT INTO ppep(seq, scrubbed, deppep_id) VALUES (?,?,?)", |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
376 (ppep, scrubbed, deppep_id) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
377 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
378 # def generate_deppep(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
379 # cur.execute("SELECT seq FROM deppep") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
380 # for row in cur.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
381 # yield row[0] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
382 cur.execute("SELECT count(*) FROM (SELECT seq FROM deppep GROUP BY seq)") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
383 for row in cur.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
384 deppep_count = row[0] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
385 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
386 cur.execute("SELECT count(*) FROM (SELECT Sequence FROM UniProtKB GROUP BY Sequence)") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
387 for row in cur.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
388 sequence_count = row[0] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
389 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
390 print( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
391 "%d phosphopeptides were read from input" % ppep_count |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
392 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
393 print( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
394 "%d corresponding dephosphopeptides are represented in input" % deppep_count |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
395 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
396 # Look for cases where both Gene_Name and Sequence are identical |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
397 cur.execute(''' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
398 SELECT Uniprot_ID, Gene_Name, Sequence |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
399 FROM UniProtKB |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
400 WHERE Sequence IN ( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
401 SELECT Sequence |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
402 FROM UniProtKB |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
403 GROUP BY Sequence, Gene_Name |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
404 HAVING count(*) > 1 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
405 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
406 ORDER BY Sequence |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
407 ''') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
408 duplicate_count = 0 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
409 old_seq = '' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
410 for row in cur.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
411 if duplicate_count == 0: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
412 print("\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).") |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
413 if row[2] != old_seq: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
414 old_seq = row[2] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
415 duplicate_count += 1 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
416 if options.warn_duplicates: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
417 print("\n%s\t%s\t%s" % row) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
418 else: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
419 if options.warn_duplicates: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
420 print("%s\t%s" % (row[0], row[1])) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
421 if duplicate_count > 0: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
422 print("\n%d sequences have duplicated accession IDs\n" % duplicate_count) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
423 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
424 print( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
425 "%s accession sequences will be searched\n" % sequence_count |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
426 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
427 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
428 #print(auto.dump()) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
429 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
430 # Convert the trie to an automaton (a finite-state machine) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
431 auto.make_automaton() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
432 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
433 # Execute query for seqs and metadata without fetching the results yet |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
434 uniprot_seq_and_id = cur.execute(UNIPROT_SEQ_AND_ID_SQL) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
435 while batch := uniprot_seq_and_id.fetchmany(size=50): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
436 if None == batch: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
437 break |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
438 for Sequence, UniProtKB_id in batch: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
439 if Sequence is not None: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
440 for end_index, (insert_order, original_value) in auto.iter(Sequence): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
441 ker.execute(''' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
442 INSERT INTO deppep_UniProtKB |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
443 (deppep_id,UniProtKB_id,pos_start,pos_end) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
444 VALUES (?,?,?,?) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
445 ''', ( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
446 insert_order, |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
447 UniProtKB_id, |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
448 1 + end_index - len(original_value), |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
449 end_index |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
450 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
451 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
452 else: |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
453 raise ValueError("UniProtKB_id %s, but Sequence is None: Check whether SwissProt file is missing sequence for this ID" % (UniProtKB_id,)) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
454 ker.execute(""" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
455 SELECT count(*) || ' accession-peptide-phosphopeptide combinations were found' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
456 FROM uniprotkb_pep_ppep_view |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
457 """ |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
458 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
459 for row in ker.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
460 print(row[0]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
461 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
462 ker.execute(""" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
463 SELECT count(*) || ' accession matches were found', count(*) AS accession_count |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
464 FROM ( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
465 SELECT accession |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
466 FROM uniprotkb_pep_ppep_view |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
467 GROUP BY accession |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
468 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
469 """ |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
470 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
471 for row in ker.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
472 print(row[0]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
473 accession_count = row[1] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
474 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
475 ker.execute(""" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
476 SELECT count(*) || ' peptide matches were found' |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
477 FROM ( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
478 SELECT peptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
479 FROM uniprotkb_pep_ppep_view |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
480 GROUP BY peptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
481 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
482 """ |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
483 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
484 for row in ker.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
485 print(row[0]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
486 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
487 ker.execute(""" |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
488 SELECT count(*) || ' phosphopeptide matches were found', count(*) AS phosphopeptide_count |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
489 FROM ( |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
490 SELECT phosphopeptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
491 FROM uniprotkb_pep_ppep_view |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
492 GROUP BY phosphopeptide |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
493 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
494 """ |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
495 ) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
496 for row in ker.fetchall(): |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
497 print(row[0]) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
498 phosphopeptide_count = row[1] |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
499 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
500 con.commit() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
501 ker.execute('vacuum') |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
502 con.close() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
503 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
504 if __name__ == "__main__": |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
505 wrap_start_time = time.perf_counter() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
506 __main__() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
507 wrap_stop_time = time.perf_counter() |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
508 # print(wrap_start_time) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
509 # print(wrap_stop_time) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
510 print("\nThe matching process took %d milliseconds to run.\n" % ((wrap_stop_time - wrap_start_time)*1000),) |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
511 |
c1403d18c189
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mqppep commit bb6c941be50db4c0719efdeaa904d7cb7aa1d182"
eschen42
parents:
diff
changeset
|
512 # vim: sw=4 ts=4 et ai : |