Mercurial > repos > jjohnson > mzsqlite_psm_align

annotate mzsqlite_psm_align.py @ 6:100cff4db790 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit ae4270ce6b1052d7eaac8e1e816263ad17f3db51-dirty
author jjohnson
date Tue, 10 Apr 2018 11:58:31 -0400
parents af5f22779a8e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
1 #!/usr/bin/env python
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
2 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
3 #
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
4 #------------------------------------------------------------------------------
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
5 # University of Minnesota
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
6 # Copyright 2017, Regents of the University of Minnesota
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
7 #------------------------------------------------------------------------------
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
8 # Author:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
9 #
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
10 # James E Johnson
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
11 #
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
12 #------------------------------------------------------------------------------
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
13 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
14
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
15 from __future__ import print_function
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
16
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
17 import argparse
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
18 import re
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
19 import sys
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
20 import sqlite3 as sqlite
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
21 from time import time
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
22
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
23
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
24 from Bio.Seq import reverse_complement, translate
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
25
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
26 ## from bedutil import bed_from_line
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
27
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
28 ## import digest
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
29
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
30 ## from ensembl_rest import get_cdna
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
31
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
32 import pysam
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
33 from twobitreader import TwoBitFile
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
34
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
35 from profmt import PROBAM_DEFAULTS,ProBAM,ProBAMEntry,ProBED,ProBEDEntry
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
36
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
37 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
38 inputs
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
39 proBed
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
40 mzIdentML
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
41 twobit
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
42 bam
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
43
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
44 inputs
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
45 mz.sqlite
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
46 genomic.mapping
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
47 bam
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
48
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
49 CREATE TABLE spectrum_identification_results (id TEXT PRIMARY KEY, spectraData_ref TEXT, spectrumID TEXT, spectrumTitle TEXT);
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
50 CREATE TABLE spectrum_identification_result_items (id TEXT PRIMARY KEY, spectrum_identification_result_ref TEXT, passThreshold TEXT, rank INTEGER, peptide_ref TEXT, calculatedMassToCharge FLOAT, experimentalMassToCharge FLOAT, chargeState INTEGER);
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
51 CREATE TABLE peptide_evidence (id TEXT PRIMARY KEY, dBSequence_ref TEXT, isDecoy TEXT, pre TEXT, post TEXT, start INTEGER, end INTEGER, peptide_ref TEXT);
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
52 CREATE TABLE db_sequence (id TEXT PRIMARY KEY , accession TEXT, searchDatabase_ref TEXT, description TEXT, sequence TEXT, length INTEGER);
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
53
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
54 SELECT
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
55 FROM spectrum_identification_result_items siri
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
56 JOIN peptide_evidence pe ON siri.peptide_ref = pe.peptide_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
57 JOIN db_sequence dbs ON pe.dBSequence_ref =
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
58 WHERE pe.isDecoy = 'false'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
59
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
60 SELECT
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
61 psm.spectrumID,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
62 psm.spectrumTitle as "QNAME",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
63 psm.id,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
64 psm.sequence,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
65 psm.passThreshold,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
66 psm."PeptideShaker PSM confidence",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
67 psm."PeptideShaker PSM score",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
68 pe.start,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
69 pe.end,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
70 pe.pre,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
71 pe.post,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
72 pe.dBSequence_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
73 FROM psm_entries psm
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
74 JOIN peptide_evidence pe ON psm.id = pe.peptide_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
75 JOIN db_sequence dbs ON pe.dBSequence_ref = dbs.accession
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
76 WHERE pe.isDecoy = 'false'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
77 AND pe.peptide_ref = 'SFYPEEVSSMVITK_15.99491461956-ATAA-10'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
78 ORDER BY psm.spectrumID
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
79
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
80
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
81 proBed to SQLite
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
82 or index proBed
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
83
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
84 for psm in psms:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
85 beds = get_bed(protein_acc)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
86 cds = ''
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
87 for bed in beds:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
88 bed.seq = twobit[bed.chrom][bed.start,bed.end]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
89 cds += bed.get_cds()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
90 refprot = translate(cds)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
91
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
92 def read_bed(path):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
93 pdict = dict()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
94 prog = re.compile('^([^\t]+\t[^\t]+\t[^\t]+\t([^\t]+)\t.*)$')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
95 with open(path,'r') as bed:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
96 for i,line in enumerate(bed):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
97 m = prog.match(line)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
98 prot = m.groups()[1]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
99 pdict[prot] = m.groups()[0]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
100 return pdict
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
101
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
102 from pyteomics import mzid
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
103 with mzid.reader(args.mzid) as mzidrdr:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
104 for psm in mzidrdr:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
105 SpectrumIdentificationItems = psm['SpectrumIdentificationItem']
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
106 for SpectrumIdentificationItem in SpectrumIdentificationItems:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
107 PeptideEvidenceRef = SpectrumIdentificationItem['PeptideEvidenceRef']
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
108 PepEvs = [r['peptideEvidence_ref'] for r in PeptideEvidenceRef]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
109 for PepEv in PepEvs:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
110 PepRef = mzidrdr[PepEv]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
111 dBSequence_ref = PepRef['dBSequence_ref']
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
112
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
113 spectrum_peptides = count(distinct sequence) FROM psm_entries WHERE
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
114
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
115 1 QNAME String Query template NAME Spectrum name * psm.spectrumTitle
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
116 2 FLAG Int Bitwise FLAG Bitwise FLAG map.strand
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
117 3 RNAME String Reference sequence NAME Reference sequence NAME * map.chrom
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
118 4 POS Int 1-based leftmost mapping POSition 1-based leftmost mapping POSition map.start
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
119 5 -MAPQ Int MAPping Quality (Phred-scaled) - 255
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
120 6 CIGAR String Extended CIGAR string (operations: MIDN) CIGAR string * map.cigar
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
121 7 -RNEXT String Mate Reference NAME ('=' if same as RNAME) - *
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
122 8 -PNEXT Int 1-Based leftmost Mate POSition - 0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
123 9 TLEN Int observed Template LENgth - 0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
124 10 SEQ String segment SEQuence Coding sequence * genomic.seq
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
125 11 -QUAL String Query QUALity (ASCII-33=Phred base quality) - *
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
126
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
127 1 QNAME psm.spectrumTitle
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
128 2 FLAG map.strand
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
129 3 RNAME map.chrom
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
130 4 POS map.start
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
131 5 -MAPQ
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
132 6 CIGAR map.cigar
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
133 7 -RNEXT
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
134 8 -PNEXT
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
135 9 -TLEN
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
136 10 SEQ genomic.seq
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
137 11 -QUAL
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
138
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
139 'NH' : 'i' genomic_locations
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
140 'XO' : 'Z'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
141 'XL' : 'i' spectrum_peptides
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
142 'XP' : 'Z' psm.sequence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
143 'YP' : 'Z' peptide_evidence.dBSequence_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
144 'XF' : 'Z' reading_frame
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
145 'XI' : 'f'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
146 'XB' : 'Z'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
147 'XR' : 'Z'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
148 'YB' : 'Z'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
149 'YA' : 'Z'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
150 'XS' : 'f'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
151 'XQ' : 'f'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
152 'XC' : 'i'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
153 'XA' : 'i'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
154 'XM' : 'Z'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
155 'XN' : 'i'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
156 'XT' : 'i'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
157 'XE' : 'i'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
158 'XG' : 'A'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
159 'XU' : 'Z'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
160
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
161 'NH' : 'i', #number of genomic locations to which the peptide sequence maps
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
162 'XO' : 'Z', #uniqueness of the peptide mapping
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
163 'XL' : 'i', #number of peptides to which the spectrum maps
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
164 'XP' : 'Z', #peptide sequence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
165 'YP' : 'Z', #Protein accession ID from the original search result
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
166 'XF' : 'Z', #Reading frame of the peptide (0, 1, 2)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
167 'XI' : 'f', #Peptide intensity
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
168 'XB' : 'Z', #massdiff; experimental mass; calculated mass massdiff can be calculated by experimental mass - calculated mass. If any number is unavailable, the value should be left blank (such as 0.01;;).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
169 'XR' : 'Z', #reference peptide sequence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
170 'YB' : 'Z', #Preceding amino acids (2 AA, B stands for before).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
171 'YA' : 'Z', #Following amino acids (2 AA, A stands for after).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
172 'XS' : 'f', #PSM score
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
173 'XQ' : 'f', #PSM FDR (i.e. q-value or 1-PEP).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
174 'XC' : 'i', #peptide charge
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
175 'XA' : 'i', #Whether the peptide is annotated 0:yes; 1:parially unknown; 2:totally unknown;
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
176 'XM' : 'Z', #Modifications
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
177 'XN' : 'i', #Number of missed cleavages in the peptide (XP)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
178 'XT' : 'i', #Enzyme specificity
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
179 'XE' : 'i', #Enzyme used in the experiment
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
180 'XG' : 'A', #Peptide type
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
181 'XU' : 'Z', #URI
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
182
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
183
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
184 Datatype Field name Description Origin
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
185 RNAME string chrom map.chrom Reference sequence chromosome
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
186 POS uint chromStart map Start position of the first DNA base
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
187 uint chromEnd map End position of the last DNA base
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
188 QNAME string name spectrum.title Unique name
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
189 uint score Score
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
190 char[1] strand + or - for strand
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
191 uint thickStart Coding region start
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
192 uint thickEnd Coding region end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
193 uint reserved Always 0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
194 int blockCount Number of blocks
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
195 int[blockCount] blockSizes Block sizes
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
196 int[blockCount] chromStarts Block starts
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
197 YP string proteinAccession Protein accession number
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
198 XP string peptideSequence Peptide sequence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
199 XO string uniqueness Peptide uniqueness
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
200 string genomeReferenceVersion Genome reference version number
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
201 XS double psmScore PSM score
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
202 XQ double fdr Estimated global false discovery rate
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
203 XM string modifications Post-translational modifications
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
204 XC int charge Charge value
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
205 XB double expMassToCharge Experimental mass to charge value
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
206 XB double calcMassToCharge Calculated mass to charge value
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
207 int psmRank Peptide-Spectrum Match rank.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
208 string datasetID Dataset Identifier
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
209 string uri Uniform Resource Identifier
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
210
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
211 XG
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
212 N Normal peptide. The peptide sequence is contained in the reference protein sequence.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
213 V Variant peptide. A single amino acid variation (SAV) is present as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
214 W Indel peptide. An insertion or deletion is present as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
215 J Novel junction peptide. A peptide that spans a novel exon-intron boundary as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
216 A Alternative junction peptide. A peptide that spans a non-canonical exon-intron boundary as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
217 M Novel exon peptide. A peptide that resides in a novel exon that is not present in the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
218 C Cross junction peptide. A peptide that spans through a splice site (partly exonic - partly intronic).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
219 E Extension peptide. A peptide that points to a non-canonical N-terminal protein extension.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
220 B 3' UTR peptide. A peptide that maps to the 3' UTR region from the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
221 O Out-of-frame peptide. A peptide that is translated from an alternative frame as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
222 T Truncation peptide. A peptide that points to a non-canonical N-terminal protein truncation.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
223 R Reverse strand peptide. A peptide that is derived from translation of the reverse strand of the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
224 I Intron peptide. A peptide that is located in an intronic region of the reference isoform.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
225 G Gene fusion peptide. An (onco-) peptide that spans two exons of different genes, through gene-fusion.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
226 D Decoy peptide. A peptide that maps to a decoy sequence from the MS-based search strategy.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
227 U Unmapped peptide. A peptide that could not be mapped to a reference sequence.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
228 X Unknown.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
229
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
230
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
231
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
232 SELECT distinct chrom, CASE WHEN strand = '+' THEN start + cds_offset - cds_start ELSE end - cds_offset - cds_start END as "pos"
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
233 FROM feature_cds_map
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
234 WHERE name = acc_name AND cds_offset >= cds_start AND cds_offset < cds_end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
235
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
236 sqlite> select * from feature_cds_map WHERE name = 'pre_STRG.28813.4_j_5350_5470';
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
237 pre_STRG.28813.4_j_5350_5470|chr7|5074750|5074857|+|0|107
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
238 pre_STRG.28813.4_j_5350_5470|chr7|5075140|5075153|+|107|120
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
239
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
240
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
241
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
242 SELECT
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
243 pe.isDecoy,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
244 pe.dBSequence_ref,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
245 pe.start,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
246 pe.end,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
247 sr.spectrumTitle,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
248 si.rank,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
249 si.chargeState,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
250 si.calculatedMassToCharge,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
251 si.experimentalMassToCharge
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
252 FROM spectrum_identification_results sr
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
253 JOIN spectrum_identification_result_items si ON si.spectrum_identification_result_ref = sr.id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
254 JOIN peptide_evidence pe ON si.peptide_ref = pe.peptide_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
255 WHERE si.id = 'SII_7389_1'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
256 ORDER BY si.rank;
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
257
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
258 SELECT pe.isDecoy, pe.dBSequence_ref, pe.start, pe.end, sr.spectrumTitle, si.rank, si.chargeState, si.calculatedMassToCharge, si.experimentalMassToCharge
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
259 FROM spectrum_identification_results sr
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
260 JOIN spectrum_identification_result_items si ON si.spectrum_identification_result_ref = sr.id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
261 JOIN peptide_evidence pe ON si.peptide_ref = pe.peptide_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
262 WHERE si.id = 'SII_7389_1'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
263 ORDER BY si.rank;
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
264
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
265
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
266
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
267 CREATE TABLE spectrum_identification_results (id TEXT PRIMARY KEY, spectraData_ref TEXT, spectrumID TEXT, spectrumTitle TEXT);
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
268 CREATE TABLE spectrum_identification_result_items (id TEXT PRIMARY KEY, spectrum_identification_result_ref TEXT, passThreshold TEXT, rank INTEGER, peptide_ref TEXT, calculatedMassToCharge FLOAT, experimentalMassToCharge FLOAT, chargeState INTEGER);
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
269 CREATE TABLE peptide_evidence (id TEXT PRIMARY KEY, dBSequence_ref TEXT, isDecoy TEXT, pre TEXT, post TEXT, start INTEGER, end INTEGER, peptide_ref TEXT);
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
270 CREATE TABLE db_sequence (id TEXT PRIMARY KEY , accession TEXT, searchDatabase_ref TEXT, description TEXT, sequence TEXT, length INTEGER);
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
271
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
272 {'write_probed': 0.08575654029846191, 'PSM_QUERY': 4.704349040985107, 'get_cds': 0.21015286445617676, 'SPECTRUM_PEPTIDES_QUERY': 32.92655086517334, 'PEPTIDE_ACC_QUERY': 425.11919951438904, 'get_mapping': 1.5911591053009033, 'GENOMIC_POS_QUERY': 10.909647226333618}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
273 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
274
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
275
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
276
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
277 def regex_match(expr, item):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
278 return re.match(expr, item) is not None
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
279
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
280
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
281 def regex_search(expr, item):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
282 return re.search(expr, item) is not None
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
283
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
284
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
285 def regex_sub(expr, replace, item):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
286 return re.sub(expr, replace, item)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
287
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
288
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
289 def get_connection(sqlitedb_path, addfunctions=True):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
290 conn = sqlite.connect(sqlitedb_path)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
291 if addfunctions:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
292 conn.create_function("re_match", 2, regex_match)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
293 conn.create_function("re_search", 2, regex_search)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
294 conn.create_function("re_sub", 3, regex_sub)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
295 return conn
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
296
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
297 PSM_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
298 SELECT
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
299 pe.dBSequence_ref,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
300 pe.start,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
301 pe.end,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
302 pe.pre,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
303 pe.post,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
304 pep.sequence,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
305 sr.id,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
306 sr.spectrumTitle,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
307 si.rank,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
308 si.chargeState,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
309 si.calculatedMassToCharge,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
310 si.experimentalMassToCharge,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
311 si.peptide_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
312 FROM spectrum_identification_results sr
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
313 JOIN spectrum_identification_result_items si ON si.spectrum_identification_result_ref = sr.id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
314 JOIN peptide_evidence pe ON si.peptide_ref = pe.peptide_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
315 JOIN peptides pep ON pe.peptide_ref = pep.id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
316 WHERE pe.isDecoy = 'false'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
317 ORDER BY sr.spectrumTitle,si.rank
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
318 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
319
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
320 PEP_MODS_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
321 SELECT location, residue, name, modType, '' as "unimod"
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
322 FROM peptide_modifications
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
323 WHERE peptide_ref = :peptide_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
324 ORDER BY location, modType, name
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
325 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
326
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
327 #number of peptides to which the spectrum maps
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
328 ## spectrum_identification_results => spectrum_identification_result_items -> peptide_evidence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
329 SPECTRUM_PEPTIDES_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
330 SELECT count(distinct pep.sequence)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
331 FROM spectrum_identification_results sr
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
332 JOIN spectrum_identification_result_items si ON si.spectrum_identification_result_ref = sr.id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
333 JOIN peptide_evidence pe ON si.peptide_ref = pe.peptide_ref
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
334 JOIN peptides pep ON pe.peptide_ref = pep.id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
335 WHERE pe.isDecoy = 'false'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
336 AND sr.id = :sr_id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
337 GROUP BY sr.id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
338 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
339 #number of genomic locations to which the peptide sequence maps
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
340 #uniqueness of the peptide mapping
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
341 ## peptides => peptide_evidence -> db_sequence -> location
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
342 ## proteins_by_peptide
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
343 PEPTIDE_ACC_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
344 SELECT
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
345 pe.dBSequence_ref,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
346 pe.start,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
347 pe.end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
348 FROM peptide_evidence pe
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
349 JOIN peptides pep ON pe.peptide_ref = pep.id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
350 WHERE pe.isDecoy = 'false'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
351 AND pep.sequence = :sequence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
352 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
353
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
354 MAP_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
355 SELECT distinct *
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
356 FROM feature_cds_map
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
357 WHERE name = :acc
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
358 AND :p_start < cds_end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
359 AND :p_end >= cds_start
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
360 ORDER BY name,cds_start,cds_end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
361 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
362
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
363 GENOMIC_POS_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
364 SELECT distinct chrom, CASE WHEN strand = '+' THEN start + :cds_offset - cds_start ELSE end - :cds_offset - cds_start END as "pos"
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
365 FROM feature_cds_map
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
366 WHERE name = :acc
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
367 AND :cds_offset >= cds_start
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
368 AND :cds_offset < cds_end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
369 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
370
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
371 FEATURE_CONTAIN_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
372 SELECT id,seqid,start,end,featuretype,strand,frame
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
373 FROM features
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
374 WHERE seqid = :seqid AND start <= :start AND end >= :end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
375 AND strand = :strand AND featuretype = :ftype
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
376 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
377
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
378 FEATURE_OVERLAP_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
379 SELECT id,seqid,start,end,featuretype,strand,frame
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
380 FROM features
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
381 WHERE seqid = :seqid
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
382 AND :end >= start AND :start <= end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
383 AND strand = :strand AND featuretype = :ftype
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
384 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
385
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
386 FEATURE_ANY_QUERY = """\
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
387 SELECT id,seqid,start,end,featuretype,strand,CAST(frame AS INTEGER) as "frame", CAST(frame AS INTEGER)==:frame as "in_frame"
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
388 FROM features
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
389 WHERE seqid = :seqid
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
390 AND :end >= start AND :start <= end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
391 AND featuretype in ('CDS','five_prime_utr','three_prime_utr','transcript')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
392 ORDER BY strand == :strand DESC, featuretype,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
393 start <= :start AND end >= :end DESC,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
394 in_frame DESC, end - start, start DESC, end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
395 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
396
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
397 def __main__():
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
398 parser = argparse.ArgumentParser(
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
399 description='Generate proBED and proBAM from mz.sqlite')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
400 parser.add_argument('mzsqlite', help="mz.sqlite converted from mzIdentML")
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
401 parser.add_argument('genomic_mapping_sqlite', help="genomic_mapping.sqlite with feature_cds_map table")
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
402 parser.add_argument(
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
403 '-R', '--genomeReference', default='Unknown',
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
404 help='Genome reference sequence in 2bit format')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
405 parser.add_argument(
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
406 '-t', '--twobit', default=None,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
407 help='Genome reference sequence in 2bit format')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
408 parser.add_argument(
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
409 '-r', '--reads_bam', default=None,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
410 help='reads alignment bam path')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
411 parser.add_argument(
5
af5f22779a8e planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit d65efea03eb2db8a43e64599a4f899ead1a252ba-dirty
jjohnson
parents: 0
diff changeset
412 '-g', '--gffutils_sqlite', default=None,
0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
413 help='gffutils GTF sqlite DB')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
414 parser.add_argument(
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
415 '-B', '--probed', default=None,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
416 help='proBed path')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
417 parser.add_argument(
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
418 '-s', '--prosam', default=None,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
419 help='proSAM path')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
420 parser.add_argument(
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
421 '-b', '--probam', default=None,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
422 help='proBAM path')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
423 parser.add_argument(
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
424 '-l', '--limit', type=int, default=None,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
425 help='limit numbers of PSMs for testing')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
426 parser.add_argument('-v', '--verbose', action='store_true', help='Verbose')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
427 parser.add_argument('-d', '--debug', action='store_true', help='Debug')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
428 args = parser.parse_args()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
429
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
430 def get_sequence(chrom, start, end):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
431 if twobit:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
432 if chrom in twobit and 0 <= start < end < len(twobit[chrom]):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
433 return twobit[chrom][start:end]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
434 contig = chrom[3:] if chrom.startswith('chr') else 'chr%s' % chrom
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
435 if contig in twobit and 0 <= start < end < len(twobit[contig]):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
436 return twobit[contig][start:end]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
437 return ''
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
438 return None
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
439
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
440 twobit = TwoBitFile(args.twobit) if args.twobit else None
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
441 samfile = pysam.AlignmentFile(args.reads_bam, "rb" ) if args.reads_bam else None
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
442 seqlens = twobit.sequence_sizes()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
443
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
444 probed = open(args.probed,'w') if args.probed else sys.stdout
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
445
5
af5f22779a8e planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit d65efea03eb2db8a43e64599a4f899ead1a252ba-dirty
jjohnson
parents: 0
diff changeset
446 gff_cursor = get_connection(args.gffutils_sqlite).cursor() if args.gffutils_sqlite else None
0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
447 map_cursor = get_connection(args.genomic_mapping_sqlite).cursor()
5
af5f22779a8e planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit d65efea03eb2db8a43e64599a4f899ead1a252ba-dirty
jjohnson
parents: 0
diff changeset
448 mz_cursor = get_connection(args.mzsqlite).cursor()
0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
449
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
450 unmapped_accs = set()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
451 timings = dict()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
452 def add_time(name,elapsed):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
453 if name in timings:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
454 timings[name] += elapsed
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
455 else:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
456 timings[name] = elapsed
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
457
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
458 XG_TYPES = ['N','V','W','J','A','M','C','E','B','O','T','R','I','G','D','U','X','*']
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
459 FT_TYPES = ['CDS','five_prime_utr','three_prime_utr','transcript']
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
460 def get_peptide_type(exons):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
461 ## XG classify peptide
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
462 ## N Normal peptide. The peptide sequence is contained in the reference protein sequence.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
463 ## V Variant peptide. A single amino acid variation (SAV) is present as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
464 ## W Indel peptide. An insertion or deletion is present as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
465 ## J Novel junction peptide. A peptide that spans a novel exon-intron boundary as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
466 ## A Alternative junction peptide. A peptide that spans a non-canonical exon-intron boundary as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
467 ## M Novel exon peptide. A peptide that resides in a novel exon that is not present in the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
468 ## C Cross junction peptide. A peptide that spans through a splice site (partly exonic - partly intronic).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
469 ## E Extension peptide. A peptide that points to a non-canonical N-terminal protein extension.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
470 ## B 3' UTR peptide. A peptide that maps to the 3' UTR region from the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
471 ## O Out-of-frame peptide. A peptide that is translated from an alternative frame as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
472 ## T Truncation peptide. A peptide that points to a non-canonical N-terminal protein truncation.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
473 ## R Reverse strand peptide. A peptide that is derived from translation of the reverse strand of the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
474 ## I Intron peptide. A peptide that is located in an intronic region of the reference isoform.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
475 ## G Gene fusion peptide. An (onco-) peptide that spans two exons of different genes, through gene-fusion.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
476 ## D Decoy peptide. A peptide that maps to a decoy sequence from the MS-based search strategy.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
477 ## U Unmapped peptide. A peptide that could not be mapped to a reference sequence.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
478 ## X Unknown.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
479
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
480 peptide_type = '*'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
481 if gff_cursor:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
482 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
483 etypes = ['*'] * len(exons)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
484 efeatures = [None] * len(exons)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
485 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
486 print('exons:%d\t%s'% (len(exons),etypes),file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
487 for i,exon in enumerate(exons):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
488 (acc,gc,gs,ge,st,cs,ce) = exon
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
489 fr = cs % 3
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
490 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
491 print('exon:\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s' % (acc,gc,gs,ge,st,cs,ce,fr),file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
492 ft_params = {"seqid" : str(gc).replace('chr',''), "start" : gs, "end" : ge, 'strand' : st, 'frame' : fr, 'ftype' : 'CDS'}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
493 features = [f for f in gff_cursor.execute(FEATURE_ANY_QUERY,ft_params)]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
494 efeatures[i] = features
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
495 for i,exon in enumerate(exons):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
496 (acc,gc,gs,ge,st,cs,ce) = exon
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
497 for f in efeatures[i]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
498 (id,seqid,start,end,featuretype,strand,frame,in_frame) = f
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
499 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
500 print('feat:\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s' % (id,seqid,start,end,featuretype,strand,frame,in_frame),file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
501 if strand == st:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
502 if start <= gs and ge <= end:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
503 if in_frame:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
504 etypes[i] = 'N'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
505 break
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
506 elif XG_TYPES.index('O') < XG_TYPES.index(etypes[i]):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
507 etypes[i] = 'O'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
508 break
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
509 else:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
510 if XG_TYPES.index('O') < XG_TYPES.index(etypes[i]):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
511 etypes[i] = 'O'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
512 peptide_type = etypes[i]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
513 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
514 add_time('pep_type',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
515 return peptide_type
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
516 def classify_exon(exon,exons,features):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
517 ## N Normal peptide. The peptide sequence is contained in the reference protein sequence.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
518 # 1 exon, contained, in_frame
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
519 # 2+ exons, contained, in_frame, on_exon_boundary
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
520 ## V Variant peptide. A single amino acid variation (SAV) is present as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
521 # 1 exon, contained, in_frame, AA_mismatch
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
522 # 2+ exons, contained, in_frame, on_exon_boundary, AA_mismatch
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
523 ## W Indel peptide. An insertion or deletion is present as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
524 # 1 exon, contained, in_frame, AA_mismatch
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
525 # 2+ exons, contained, in_frame, on_exon_boundary or off by 3, AA_mismatch
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
526 ## J Novel junction peptide. A peptide that spans a novel exon-intron boundary as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
527 # 2+ exons, contained, on_exon_boundary, same transcript, non adjacent exons
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
528 ## A Alternative junction peptide. A peptide that spans a non-canonical exon-intron boundary as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
529 # 2+ exons, contained, on_exon_boundary, same transcript, non adjacent exons
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
530 ## M Novel exon peptide. A peptide that resides in a novel exon that is not present in the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
531 ## C Cross junction peptide. A peptide that spans through a splice site (partly exonic - partly intronic).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
532 # 1 exon overlaps but not contained
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
533 ## E Extension peptide. A peptide that points to a non-canonical N-terminal protein extension.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
534 ## B 3' UTR peptide. A peptide that maps to the 3' UTR region from the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
535 # exon overlaps a three_prime_utr
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
536 ## O Out-of-frame peptide. A peptide that is translated from an alternative frame as compared to the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
537 # exon contained but not in_frame
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
538 ## T Truncation peptide. A peptide that points to a non-canonical N-terminal protein truncation.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
539 ## R Reverse strand peptide. A peptide that is derived from translation of the reverse strand of the reference.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
540 ## I Intron peptide. A peptide that is located in an intronic region of the reference isoform.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
541 # exon contained in transcript, not not overlapping any exon
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
542 ## G Gene fusion peptide. An (onco-) peptide that spans two exons of different genes, through gene-fusion.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
543 # exonis from different seqs, strand, or transcripts
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
544 ## D Decoy peptide. A peptide that maps to a decoy sequence from the MS-based search strategy.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
545 ## U Unmapped peptide. A peptide that could not be mapped to a reference sequence.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
546 ## X Unknown.
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
547 return '*'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
548
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
549 def get_variant_cds(exons,ref_prot,peptide,pep_cds):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
550 if ref_prot != peptide and samfile:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
551 try:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
552 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
553 print('name: %s \nref: %s\npep: %s\n' % (scan_name,ref_prot,peptide), file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
554 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
555 for exon in exons:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
556 (acc,chrom,start,end,strand,c_start,c_end) = exon
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
557 a_start = c_start / 3 * 3
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
558 a_end = c_end / 3 * 3
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
559 if ref_prot[a_start:a_end] != peptide[a_start:a_end]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
560 pileup = get_exon_pileup(chrom,start,end)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
561 for i, (bi,ai,ao) in enumerate([(i,i / 3, i % 3) for i in range(c_start, c_end)]):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
562 if ao == 0 or i == 0:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
563 if ref_prot[ai] != peptide[ai]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
564 codon = get_pep_codon(pileup, bi - c_start, peptide[ai], ao)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
565 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
566 print('%d %d %d %s : %s %s %s' % (bi,ai,ao, peptide[ai], str(pep_cds[:bi]), str(codon), str(pep_cds[bi+3:])), file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
567 if codon:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
568 pep_cds = pep_cds[:bi] + codon + pep_cds[bi+3:]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
569 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
570 add_time('var_cds',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
571 except Exception as e:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
572 print('name: %s \nref: %s\npep: %s\n%s\n' % (scan_name,ref_prot,peptide,e), file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
573 return pep_cds
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
574
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
575 def get_mapping(acc,pep_start,pep_end):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
576 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
577 p_start = (pep_start - 1) * 3
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
578 p_end = pep_end * 3
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
579 map_params = {"acc" : acc, "p_start" : p_start, "p_end" : p_end}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
580 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
581 print('%s' % map_params, file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
582 locs = [l for l in map_cursor.execute(MAP_QUERY,map_params)]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
583 exons = []
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
584 ## ========= pep
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
585 ## --- continue
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
586 ## --- trim
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
587 ## --- copy
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
588 ## --- trim
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
589 ## --- break
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
590 c_end = 0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
591 for i, (acc,chrom,start,end,strand,cds_start,cds_end) in enumerate(locs):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
592 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
593 print('Prot: %s\t%s:%d-%d\t%s\t%d\t%d' % (acc,chrom,start,end,strand,cds_start,cds_end),file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
594 c_start = c_end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
595 if cds_end < p_start:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
596 continue
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
597 if cds_start >= p_end:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
598 break
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
599 if strand == '+':
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
600 if cds_start < p_start:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
601 start += p_start - cds_start
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
602 if cds_end > p_end:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
603 end -= cds_end - p_end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
604 else:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
605 if cds_start < p_start:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
606 end -= p_start - cds_start
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
607 if cds_end > p_end:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
608 start += cds_end - p_end
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
609 c_end = c_start + abs(end - start)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
610 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
611 print('Pep: %s\t%s:%d-%d\t%s\t%d\t%d' % (acc,chrom,start,end,strand,cds_start,cds_end),file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
612 exons.append([acc,chrom,start,end,strand,c_start,c_end])
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
613 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
614 add_time('get_mapping',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
615 return exons
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
616
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
617 def get_cds(exons):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
618 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
619 seqs = []
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
620 for i, (acc,chrom,start,end,strand,cds_start,cds_end) in enumerate(exons):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
621 seq = get_sequence(chrom, min(start,end), max(start,end))
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
622 if strand == '-':
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
623 seq = reverse_complement(seq)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
624 seqs.append(seq)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
625 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
626 add_time('get_cds',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
627 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
628 print('CDS: %s' % str(seqs),file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
629 return ''.join(seqs) if seqs else ''
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
630
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
631 def genomic_mapping_count(peptide):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
632 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
633 params = {"sequence" : peptide}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
634 acc_locs = [l for l in mz_cursor.execute(PEPTIDE_ACC_QUERY,params)]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
635 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
636 add_time('PEPTIDE_ACC_QUERY',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
637 if acc_locs:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
638 if len(acc_locs) == 1:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
639 return 1
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
640 locations = set()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
641 for i,acc_loc in enumerate(acc_locs):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
642 (acc,pep_start,pep_end) = acc_loc
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
643 if acc in unmapped_accs:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
644 continue
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
645 try:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
646 add_time('GENOMIC_POS_QUERY_COUNT',1)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
647 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
648 p_start = pep_start * 3
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
649 p_end = pep_end * 3
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
650 params = {"acc" : acc, "cds_offset" : p_start}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
651 (start_chrom,start_pos) = map_cursor.execute(GENOMIC_POS_QUERY, params).fetchone()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
652 params = {"acc" : acc, "cds_offset" : p_end}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
653 (end_chrom,end_pos) = map_cursor.execute(GENOMIC_POS_QUERY, params).fetchone()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
654 locations.add('%s:%s-%s:%s' % (start_chrom,start_pos,end_chrom,end_pos))
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
655 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
656 add_time('GENOMIC_POS_QUERY',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
657 except:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
658 unmapped_accs.add(acc)
5
af5f22779a8e planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit d65efea03eb2db8a43e64599a4f899ead1a252ba-dirty
jjohnson
parents: 0
diff changeset
659 if args.debug:
af5f22779a8e planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit d65efea03eb2db8a43e64599a4f899ead1a252ba-dirty
jjohnson
parents: 0
diff changeset
660 print('Unmapped: %s' % acc, file=sys.stderr)
0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
661 return len(locations)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
662 return -1
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
663
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
664 def spectrum_peptide_count(spectrum_id):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
665 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
666 params = {"sr_id" : spectrum_id}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
667 pep_count = mz_cursor.execute(SPECTRUM_PEPTIDES_QUERY, params).fetchone()[0]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
668 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
669 add_time('SPECTRUM_PEPTIDES_QUERY',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
670 return pep_count
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
671
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
672 def get_exon_pileup(chrom,chromStart,chromEnd):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
673 cols = []
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
674 for pileupcolumn in samfile.pileup(chrom, chromStart, chromEnd):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
675 if chromStart <= pileupcolumn.reference_pos <= chromEnd:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
676 bases = dict()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
677 col = {'depth' : 0, 'cov' : pileupcolumn.nsegments, 'pos': pileupcolumn.reference_pos, 'bases' : bases}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
678 for pileupread in pileupcolumn.pileups:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
679 if not pileupread.is_del and not pileupread.is_refskip:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
680 col['depth'] += 1
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
681 base = pileupread.alignment.query_sequence[pileupread.query_position]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
682 if base not in bases:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
683 bases[base] = 1
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
684 else:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
685 bases[base] += 1
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
686 cols.append(col)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
687 return cols
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
688
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
689 codon_map = {"TTT":"F", "TTC":"F", "TTA":"L", "TTG":"L",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
690 "TCT":"S", "TCC":"S", "TCA":"S", "TCG":"S",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
691 "TAT":"Y", "TAC":"Y", "TAA":"*", "TAG":"*",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
692 "TGT":"C", "TGC":"C", "TGA":"*", "TGG":"W",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
693 "CTT":"L", "CTC":"L", "CTA":"L", "CTG":"L",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
694 "CCT":"P", "CCC":"P", "CCA":"P", "CCG":"P",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
695 "CAT":"H", "CAC":"H", "CAA":"Q", "CAG":"Q",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
696 "CGT":"R", "CGC":"R", "CGA":"R", "CGG":"R",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
697 "ATT":"I", "ATC":"I", "ATA":"I", "ATG":"M",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
698 "ACT":"T", "ACC":"T", "ACA":"T", "ACG":"T",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
699 "AAT":"N", "AAC":"N", "AAA":"K", "AAG":"K",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
700 "AGT":"S", "AGC":"S", "AGA":"R", "AGG":"R",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
701 "GTT":"V", "GTC":"V", "GTA":"V", "GTG":"V",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
702 "GCT":"A", "GCC":"A", "GCA":"A", "GCG":"A",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
703 "GAT":"D", "GAC":"D", "GAA":"E", "GAG":"E",
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
704 "GGT":"G", "GGC":"G", "GGA":"G", "GGG":"G",}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
705
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
706 aa_codon_map = dict()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
707 for c,a in codon_map.items():
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
708 aa_codon_map[a] = [c] if a not in aa_codon_map else aa_codon_map[a] + [c]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
709
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
710 aa_na_map = dict() # m[aa]{bo : {b1 : [b3]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
711 for c,a in codon_map.items():
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
712 if a not in aa_na_map:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
713 aa_na_map[a] = dict()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
714 d = aa_na_map[a]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
715 for i in range(3):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
716 b = c[i]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
717 if i < 2:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
718 if b not in d:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
719 d[b] = dict() if i < 1 else set()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
720 d = d[b]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
721 else:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
722 d.add(b)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
723
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
724 def get_pep_codon(pileup, idx, aa, ao):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
725 try:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
726 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
727 bases = []
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
728 for i in range(3):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
729 if i < ao:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
730 bases.append(list(set([c[i] for c in aa_codon_map[aa]])))
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
731 else:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
732 bases.append([b for b, cnt in reversed(sorted(pileup[idx + i]['bases'].iteritems(), key=lambda (k,v): (v,k)))])
5
af5f22779a8e planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit d65efea03eb2db8a43e64599a4f899ead1a252ba-dirty
jjohnson
parents: 0
diff changeset
733 if args.debug:
af5f22779a8e planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit d65efea03eb2db8a43e64599a4f899ead1a252ba-dirty
jjohnson
parents: 0
diff changeset
734 print('%s' % bases,file=sys.stderr)
0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
735 for b0 in bases[0]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
736 if b0 not in aa_na_map[aa]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
737 continue
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
738 for b1 in bases[1]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
739 if b1 not in aa_na_map[aa][b0]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
740 continue
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
741 for b2 in bases[2]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
742 if b2 in aa_na_map[aa][b0][b1]:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
743 return '%s%s%s' % (b0,b1,b2)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
744 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
745 add_time('pep_codon',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
746 except Exception as e:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
747 print("get_pep_codon: %s %s %s %s"
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
748 % (aa, ao, idx, pileup), file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
749 raise e
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
750 return None
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
751
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
752 def write_probed(chrom,chromStart,chromEnd,strand,blockCount,blockSizes,blockStarts,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
753 spectrum,protacc,peptide,uniqueness,genomeReference,score=1000,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
754 psmScore='.', fdr='.', mods='.', charge='.',
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
755 expMassToCharge='.', calcMassToCharge='.',
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
756 psmRank='.', datasetID='.', uri='.'):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
757 probed.write('%s\t%d\t%d\t%s\t%d\t%s\t%d\t%d\t%s\t%d\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\n' % \
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
758 (chrom,chromStart,chromEnd,spectrum,score,strand,chromStart,chromEnd,'0',blockCount,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
759 ','.join([str(v) for v in blockSizes]),
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
760 ','.join([str(v) for v in blockStarts]),
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
761 protacc,peptide,uniqueness, genomeReference,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
762 psmScore, fdr, mods, charge, expMassToCharge, calcMassToCharge, psmRank, datasetID, uri))
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
763
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
764 def get_genomic_location(exons):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
765 chrom = exons[0][1]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
766 strand = exons[0][4]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
767 pos = [exon[2] for exon in exons] + [exon[3] for exon in exons]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
768 chromStart = min(pos)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
769 chromEnd = max(pos)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
770 blockCount = len(exons)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
771 blockSizes = [abs(exon[3] - exon[2]) for exon in exons]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
772 blockStarts = [min(exon[2],exon[3]) - chromStart for exon in exons]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
773 return (chrom,chromStart,chromEnd,strand,blockCount,blockSizes,blockStarts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
774
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
775 def get_psm_modifications(peptide_ref):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
776 mods = []
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
777 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
778 params = {"peptide_ref" : peptide_ref}
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
779 pepmods = [m for m in mz_cursor.execute(PEP_MODS_QUERY, params)]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
780 if pepmods:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
781 for (location, residue, name, modType, unimod) in pepmods:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
782 mods.append('%s-%s' % (location, unimod if unimod else '%s%s' % (name,residue)))
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
783 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
784 add_time('PEP_MODS_QUERY',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
785 return ';'.join(mods)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
786
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
787
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
788 """
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
789 QNAME
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
790 FLAG
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
791 RNAME
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
792 POS
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
793 CIGAR
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
794 SEQ
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
795 'NH' : 'i', #number of genomic locations to which the peptide sequence maps
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
796 'XO' : 'Z', #uniqueness of the peptide mapping
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
797 'XL' : 'i', #number of peptides to which the spectrum maps
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
798 'XP' : 'Z', #peptide sequence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
799 'YP' : 'Z', #Protein accession ID from the original search result
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
800 'XF' : 'Z', #Reading frame of the peptide (0, 1, 2)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
801 'XI' : 'f', #Peptide intensity
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
802 'XB' : 'Z', #massdiff; experimental mass; calculated mass massdiff can be calculated by experimental mass - calculated mass. If any number is unavailable, the value should be left blank (such as 0.01;;).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
803 'XR' : 'Z', #reference peptide sequence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
804 'YB' : 'Z', #Preceding amino acids (2 AA, B stands for before).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
805 'YA' : 'Z', #Following amino acids (2 AA, A stands for after).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
806 'XS' : 'f', #PSM score
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
807 'XQ' : 'f', #PSM FDR (i.e. q-value or 1-PEP).
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
808 'XC' : 'i', #peptide charge
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
809 'XA' : 'i', #Whether the peptide is annotated 0:yes; 1:parially unknown; 2:totally unknown;
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
810 'XM' : 'Z', #Modifications
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
811 'XN' : 'i', #Number of missed cleavages in the peptide (XP)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
812 'XT' : 'i', #Enzyme specificity
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
813 'XE' : 'i', #Enzyme used in the experiment
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
814 'XG' : 'A', #Peptide type
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
815 'XU' : 'Z', #URI
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
816 """
5
af5f22779a8e planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit d65efea03eb2db8a43e64599a4f899ead1a252ba-dirty
jjohnson
parents: 0
diff changeset
817 psm_cursor = get_connection(args.mzsqlite).cursor()
0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
818 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
819 psms = psm_cursor.execute(PSM_QUERY)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
820 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
821 add_time('PSM_QUERY',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
822 proBAM = ProBAM(species=None,assembly=args.genomeReference,seqlens=seqlens,comments=[])
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
823 proBED = ProBED(species=None,assembly=args.genomeReference,comments=[])
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
824 for i, psm in enumerate(psms):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
825 probam_dict = PROBAM_DEFAULTS.copy()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
826 (acc,pep_start,pep_end,aa_pre,aa_post,peptide,spectrum_id,spectrum_title,rank,charge,calcmass,exprmass,pepref) = psm
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
827 scan_name = spectrum_title if spectrum_title else spectrum_id
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
828 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
829 print('\nPSM: %d\t%s' % (i, '\t'.join([str(v) for v in (acc,pep_start,pep_end,peptide,spectrum_id,scan_name,rank,charge,calcmass,exprmass)])), file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
830 exons = get_mapping(acc,pep_start,pep_end)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
831 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
832 print('%s' % exons, file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
833 if not exons:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
834 continue
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
835 mods = get_psm_modifications(pepref)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
836 (chrom,chromStart,chromEnd,strand,blockCount,blockSizes,blockStarts) = get_genomic_location(exons)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
837 ref_cds = get_cds(exons)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
838 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
839 print('%s' % ref_cds, file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
840 ref_prot = translate(ref_cds)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
841 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
842 print('%s' % ref_prot, file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
843 print('%s' % peptide, file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
844 spectrum_peptides = spectrum_peptide_count(spectrum_id)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
845 peptide_locations = genomic_mapping_count(peptide)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
846 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
847 print('spectrum_peptide_count: %d\tpeptide_location_count: %d' % (spectrum_peptides,peptide_locations), file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
848 uniqueness = 'unique' if peptide_locations == 1 else 'not-unique[unknown]'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
849 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
850 proBEDEntry = ProBEDEntry(chrom,chromStart,chromEnd,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
851 '%s_%s' % (acc,scan_name),
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
852 1000,strand,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
853 blockCount,blockSizes,blockStarts,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
854 acc,peptide,uniqueness,args.genomeReference,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
855 charge=charge,expMassToCharge=exprmass,calcMassToCharge=calcmass,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
856 mods=mods if mods else '.', psmRank=rank)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
857 proBED.add_entry(proBEDEntry)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
858 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
859 add_time('add_probed',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
860 if len(ref_prot) != len(peptide):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
861 continue
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
862 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
863 probam_dict['NH'] = peptide_locations
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
864 probam_dict['XO'] = uniqueness
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
865 probam_dict['XL'] = peptide_locations
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
866 probam_dict['XP'] = peptide
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
867 probam_dict['YP'] = acc
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
868 probam_dict['XC'] = charge
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
869 probam_dict['XB'] = '%f;%f;%f' % (exprmass - calcmass, exprmass, calcmass)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
870 probam_dict['XR'] = ref_prot # ? dbSequence
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
871 probam_dict['YA'] = aa_post
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
872 probam_dict['YB'] = aa_pre
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
873 probam_dict['XM'] = mods if mods else '*'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
874 flag = 16 if strand == '-' else 0
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
875 if str(rank)!=str(1) and rank!='*' and rank!=[] and rank!="":
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
876 flag += 256
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
877 probam_dict['XF'] = ','.join([str(e[2] % 3) for e in exons])
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
878 ## check for variation from ref_cds
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
879 pep_cds = get_variant_cds(exons,ref_prot,peptide,ref_cds)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
880 peptide_type = '*'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
881 ## XG classify peptide
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
882 probam_dict['XG'] = get_peptide_type(exons)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
883 ## probam_dict['MD'] = peptide
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
884
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
885 ## FIX SAM sequence is forward strand
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
886 seq = pep_cds if strand == '+' else reverse_complement(pep_cds)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
887 ## cigar based on plus strand
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
888 cigar = ''
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
889 if strand == '+':
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
890 blkStarts = blockStarts
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
891 blkSizes = blockSizes
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
892 else:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
893 blkStarts = [x for x in reversed(blockStarts)]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
894 blkSizes = [x for x in reversed(blockSizes)]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
895 for j in range(blockCount):
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
896 if j > 0:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
897 intron = blkStarts[j] - (blkStarts[j-1] + blkSizes[j-1])
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
898 if intron > 0:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
899 cigar += '%dN' % intron
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
900 cigar += '%dM' % blkSizes[j]
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
901 ## Mods TODO
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
902 proBAMEntry = ProBAMEntry(qname=scan_name, flag=flag, rname=chrom, pos=chromStart+1,
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
903 cigar=cigar,seq=seq,optional=probam_dict)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
904 proBAM.add_entry(proBAMEntry)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
905 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
906 add_time('add_probam',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
907
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
908 if args.debug:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
909 print('%s' % probam_dict, file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
910
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
911 if args.limit and i >= args.limit:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
912 break
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
913 if args.probed:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
914 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
915 with open(args.probed,'w') as fh:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
916 proBED.write(fh)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
917 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
918 add_time('write_probed',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
919 if args.prosam or args.probam:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
920 samfile = args.prosam if args.prosam else 'temp.sam'
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
921 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
922 with open(samfile,'w') as fh:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
923 proBAM.write(fh)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
924 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
925 add_time('write_prosam',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
926 if args.probam:
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
927 ts = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
928 bamfile = args.prosam.replace('.sam','.bam')
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
929 pysam.view(samfile, '-b', '-o', args.probam, catch_stdout=False)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
930 te = time()
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
931 add_time('write_probam',te - ts)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
932 pysam.index(args.probam)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
933
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
934 print('\n%s\n' % str(timings), file=sys.stderr)
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
935
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
936 if __name__ == "__main__":
492f98d89e26 planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/mzsqlite_psm_align commit 88e2fb9c31fbd687a0956924a870137d1fb9bee3-dirty
jjohnson
parents:
diff changeset
937 __main__()