Mercurial > repos > tomnl > metfrag
annotate metfrag.py @ 2:d040e27b6225 draft
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
| author | tomnl |
|---|---|
| date | Mon, 03 Jun 2019 11:20:55 -0400 |
| parents | c1b168770b68 |
| children | 5ee936e570a7 |
| rev | line source |
|---|---|
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
1 from __future__ import absolute_import, print_function |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
2 import argparse |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
3 import csv |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
4 import os |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
5 import sys |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
6 import six |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
7 import re |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
8 import random |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
9 import string |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
10 import shutil |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
11 import glob |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
12 import tempfile |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
13 import multiprocessing |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
14 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
15 from subprocess import call |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
16 from collections import defaultdict |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
17 |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
18 print(sys.version) |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
19 |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
20 parser = argparse.ArgumentParser() |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
21 parser.add_argument('--input_pth') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
22 parser.add_argument('--result_pth', default='metfrag_result.csv') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
23 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
24 parser.add_argument('--temp_dir') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
25 parser.add_argument('--polarity', default='pos') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
26 parser.add_argument('--minMSMSpeaks', default=1) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
27 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
28 parser.add_argument('--MetFragDatabaseType', default='PubChem') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
29 parser.add_argument('--LocalDatabasePath', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
30 parser.add_argument('--LocalMetChemDatabaseServerIp', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
31 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
32 parser.add_argument('--DatabaseSearchRelativeMassDeviation', default=5) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
33 parser.add_argument('--FragmentPeakMatchRelativeMassDeviation', default=10) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
34 parser.add_argument('--FragmentPeakMatchAbsoluteMassDeviation', default=0.001) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
35 parser.add_argument('--NumberThreads', default=1) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
36 parser.add_argument('--UnconnectedCompoundFilter', action='store_true') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
37 parser.add_argument('--IsotopeFilter', action='store_true') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
38 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
39 parser.add_argument('--FilterMinimumElements', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
40 parser.add_argument('--FilterMaximumElements', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
41 parser.add_argument('--FilterSmartsInclusionList', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
42 parser.add_argument('--FilterSmartsExclusionList', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
43 parser.add_argument('--FilterIncludedElements', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
44 parser.add_argument('--FilterExcludedElements', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
45 parser.add_argument('--FilterIncludedExclusiveElements', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
46 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
47 parser.add_argument('--score_thrshld', default=0) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
48 parser.add_argument('--pctexplpeak_thrshld', default=0) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
49 parser.add_argument('--schema') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
50 parser.add_argument('--cores_top_level', default=1) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
51 parser.add_argument('--chunks', default=1) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
52 parser.add_argument('--meta_select_col', default='name') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
53 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
54 parser.add_argument('--ScoreSuspectLists', default='') |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
55 parser.add_argument('--MetFragScoreTypes', default="FragmenterScore,OfflineMetFusionScore") |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
56 parser.add_argument('--MetFragScoreWeights', default="1.0,1.0") |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
57 |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
58 args = parser.parse_args() |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
59 print(args) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
60 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
61 # Create temporary working directory |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
62 if args.temp_dir: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
63 wd = args.temp_dir |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
64 else: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
65 wd = tempfile.mkdtemp() |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
66 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
67 if os.path.exists(wd): |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
68 shutil.rmtree(wd) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
69 os.makedirs(wd) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
70 else: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
71 os.makedirs(wd) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
72 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
73 ###################################################################### |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
74 # Setup regular expressions for MSP parsing dictionary |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
75 ###################################################################### |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
76 regex_msp = {} |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
77 regex_msp['name'] = ['^Name(?:=|:)(.*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
78 regex_msp['polarity'] = ['^ion.*mode(?:=|:)(.*)$', '^ionization.*mode(?:=|:)(.*)$', '^polarity(?:=|:)(.*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
79 regex_msp['precursor_mz'] = ['^precursor.*m/z(?:=|:)\s*(\d*[.,]?\d*)$', '^precursor.*mz(?:=|:)\s*(\d*[.,]?\d*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
80 regex_msp['precursor_type'] = ['^precursor.*type(?:=|:)(.*)$', '^adduct(?:=|:)(.*)$', '^ADDUCTIONNAME(?:=|:)(.*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
81 regex_msp['num_peaks'] = ['^Num.*Peaks(?:=|:)\s*(\d*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
82 regex_msp['msp'] = ['^Name(?:=|:)(.*)$'] # Flag for standard MSP format |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
83 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
84 regex_massbank = {} |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
85 regex_massbank['name'] = ['^RECORD_TITLE:(.*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
86 regex_massbank['polarity'] = ['^AC\$MASS_SPECTROMETRY:\s+ION_MODE\s+(.*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
87 regex_massbank['precursor_mz'] = ['^MS\$FOCUSED_ION:\s+PRECURSOR_M/Z\s+(\d*[.,]?\d*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
88 regex_massbank['precursor_type'] = ['^MS\$FOCUSED_ION:\s+PRECURSOR_TYPE\s+(.*)$'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
89 regex_massbank['num_peaks'] = ['^PK\$NUM_PEAK:\s+(\d*)'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
90 regex_massbank['cols'] = ['^PK\$PEAK:\s+(.*)'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
91 regex_massbank['massbank'] = ['^RECORD_TITLE:(.*)$'] # Flag for massbank format |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
92 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
93 if args.schema == 'msp': |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
94 meta_regex = regex_msp |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
95 elif args.schema == 'massbank': |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
96 meta_regex = regex_massbank |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
97 elif args.schema == 'auto': |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
98 # If auto we just check for all the available paramter names and then determine if Massbank or MSP based on |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
99 # the name parameter |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
100 meta_regex = {} |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
101 meta_regex.update(regex_massbank) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
102 meta_regex['name'].extend(regex_msp['name']) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
103 meta_regex['polarity'].extend(regex_msp['polarity']) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
104 meta_regex['precursor_mz'].extend(regex_msp['precursor_mz']) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
105 meta_regex['precursor_type'].extend(regex_msp['precursor_type']) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
106 meta_regex['num_peaks'].extend(regex_msp['num_peaks']) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
107 meta_regex['msp'] = regex_msp['msp'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
108 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
109 print(meta_regex) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
110 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
111 adduct_types = { |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
112 '[M+H]+': 1.007276, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
113 '[M+NH4]+': 18.034374, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
114 '[M+Na]+': 22.989218, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
115 '[M+K]+': 38.963158, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
116 '[M+CH3OH+H]+': 33.033489, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
117 '[M+ACN+H]+': 42.033823, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
118 '[M+ACN+Na]+': 64.015765, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
119 '[M+2ACN+H]+': 83.06037, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
120 '[M-H]-': -1.007276, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
121 '[M+Cl]-': 34.969402, |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
122 } |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
123 |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
124 # function to extract the meta data using the regular expressions |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
125 def parse_meta(meta_regex, meta_info={}): |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
126 for k, regexes in six.iteritems(meta_regex): |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
127 for reg in regexes: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
128 m = re.search(reg, line, re.IGNORECASE) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
129 if m: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
130 meta_info[k] = '-'.join(m.groups()).strip() |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
131 return meta_info |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
132 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
133 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
134 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
135 ###################################################################### |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
136 # Setup parameter dictionary |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
137 ###################################################################### |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
138 def init_paramd(args): |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
139 paramd = defaultdict() |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
140 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
141 paramd["MetFragDatabaseType"] = args.MetFragDatabaseType |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
142 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
143 if args.MetFragDatabaseType == "LocalCSV": |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
144 paramd["LocalDatabasePath"] = args.LocalDatabasePath |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
145 elif args.MetFragDatabaseType == "MetChem": |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
146 paramd["LocalMetChemDatabase"] = "metchem" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
147 paramd["LocalMetChemDatabasePortNumber"] = 5432 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
148 paramd["LocalMetChemDatabaseServerIp"] = args.LocalMetChemDatabaseServerIp |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
149 paramd["LocalMetChemDatabaseUser"] = "metchemro" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
150 paramd["LocalMetChemDatabasePassword"] = "metchemro" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
151 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
152 paramd["FragmentPeakMatchAbsoluteMassDeviation"] = args.FragmentPeakMatchAbsoluteMassDeviation |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
153 paramd["FragmentPeakMatchRelativeMassDeviation"] = args.FragmentPeakMatchRelativeMassDeviation |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
154 paramd["DatabaseSearchRelativeMassDeviation"] = args.DatabaseSearchRelativeMassDeviation |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
155 paramd["SampleName"] = '' |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
156 paramd["ResultsPath"] = os.path.join(wd) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
157 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
158 if args.polarity == "pos": |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
159 paramd["IsPositiveIonMode"] = True |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
160 paramd["PrecursorIonModeDefault"] = "1" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
161 paramd["PrecursorIonMode"] = "1" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
162 paramd["nm_mass_diff_default"] = 1.007276 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
163 else: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
164 paramd["IsPositiveIonMode"] = False |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
165 paramd["PrecursorIonModeDefault"] = "-1" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
166 paramd["PrecursorIonMode"] = "-1" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
167 paramd["nm_mass_diff_default"] = -1.007276 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
168 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
169 paramd["MetFragCandidateWriter"] = "CSV" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
170 paramd["NumberThreads"] = args.NumberThreads |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
171 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
172 if args.ScoreSuspectLists: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
173 paramd["ScoreSuspectLists"] = args.ScoreSuspectLists |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
174 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
175 paramd["MetFragScoreTypes"] = args.MetFragScoreTypes |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
176 paramd["MetFragScoreWeights"] = args.MetFragScoreWeights |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
177 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
178 dct_filter = defaultdict() |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
179 filterh = [] |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
180 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
181 if args.UnconnectedCompoundFilter: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
182 filterh.append('UnconnectedCompoundFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
183 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
184 if args.IsotopeFilter: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
185 filterh.append('IsotopeFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
186 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
187 if args.FilterMinimumElements: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
188 filterh.append('MinimumElementsFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
189 dct_filter['FilterMinimumElements'] = args.FilterMinimumElements |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
190 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
191 if args.FilterMaximumElements: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
192 filterh.append('MaximumElementsFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
193 dct_filter['FilterMaximumElements'] = args.FilterMaximumElements |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
194 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
195 if args.FilterSmartsInclusionList: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
196 filterh.append('SmartsSubstructureInclusionFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
197 dct_filter['FilterSmartsInclusionList'] = args.FilterSmartsInclusionList |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
198 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
199 if args.FilterSmartsExclusionList: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
200 filterh.append('SmartsSubstructureExclusionFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
201 dct_filter['FilterSmartsExclusionList'] = args.FilterSmartsExclusionList |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
202 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
203 # My understanding is that both 'ElementInclusionExclusiveFilter' and 'ElementExclusionFilter' use |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
204 # 'FilterIncludedElements' |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
205 if args.FilterIncludedExclusiveElements: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
206 filterh.append('ElementInclusionExclusiveFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
207 dct_filter['FilterIncludedElements'] = args.FilterIncludedExclusiveElements |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
208 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
209 if args.FilterIncludedElements: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
210 filterh.append('ElementInclusionFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
211 dct_filter['FilterIncludedElements'] = args.FilterIncludedElements |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
212 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
213 if args.FilterExcludedElements: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
214 filterh.append('ElementExclusionFilter') |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
215 dct_filter['FilterExcludedElements'] = args.FilterExcludedElements |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
216 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
217 if filterh: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
218 fcmds = ','.join(filterh) + ' ' |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
219 for k, v in six.iteritems(dct_filter): |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
220 fcmds += "{0}={1} ".format(str(k), str(v)) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
221 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
222 paramd["MetFragPreProcessingCandidateFilter"] = fcmds |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
223 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
224 return paramd |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
225 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
226 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
227 ###################################################################### |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
228 # Function to run metfrag when all metainfo and peaks have been parsed |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
229 ###################################################################### |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
230 def run_metfrag(meta_info, peaklist, args, wd, spectrac, adduct_types): |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
231 # Get sample details (if possible to extract) e.g. if created as part of the msPurity pipeline) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
232 # choose between getting additional details to add as columns as either all meta data from msp, just |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
233 # details from the record name (i.e. when using msPurity and we have the columns coded into the name) or |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
234 # just the spectra index (spectrac)] |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
235 # Returns the parameters used and the command line call |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
236 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
237 paramd = init_paramd(args) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
238 if args.meta_select_col == 'name': |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
239 # have additional column of just the name |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
240 paramd['additional_details'] = {'name': meta_info['name']} |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
241 elif args.meta_select_col == 'name_split': |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
242 # have additional columns split by "|" and then on ":" e.g. MZ:100.2 | RT:20 | xcms_grp_id:1 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
243 paramd['additional_details'] = {sm.split(":")[0].strip(): sm.split(":")[1].strip() for sm in |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
244 meta_info['name'].split("|")} |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
245 elif args.meta_select_col == 'all': |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
246 # have additional columns based on all the meta information extracted from the MSP |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
247 paramd['additional_details'] = meta_info |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
248 else: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
249 # Just have and index of the spectra in the MSP file |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
250 paramd['additional_details'] = {'spectra_idx': spectrac} |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
251 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
252 paramd["SampleName"] = "{}_metfrag_result".format(spectrac) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
253 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
254 # =============== Output peaks to txt file ============================== |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
255 paramd["PeakListPath"] = os.path.join(wd, "{}_tmpspec.txt".format(spectrac)) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
256 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
257 # write spec file |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
258 with open(paramd["PeakListPath"], 'w') as outfile: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
259 for p in peaklist: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
260 outfile.write(p[0] + "\t" + p[1] + "\n") |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
261 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
262 # =============== Update param based on MSP metadata ====================== |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
263 # Replace param details with details from MSP if required |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
264 if 'precursor_type' in meta_info and meta_info['precursor_type'] in adduct_types: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
265 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
266 nm = float(meta_info['precursor_mz']) + adduct_types[meta_info['precursor_type']] |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
267 paramd["PrecursorIonMode"] = int(round(adduct_types[meta_info['precursor_type']], 0)) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
268 else: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
269 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
270 paramd["PrecursorIonMode"] = paramd['PrecursorIonModeDefault'] |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
271 nm = float(meta_info['precursor_mz']) + paramd['nm_mass_diff_default'] |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
272 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
273 paramd["NeutralPrecursorMass"] = nm |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
274 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
275 # =============== Create CLI cmd for metfrag =============================== |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
276 cmd = "metfrag" |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
277 for k, v in six.iteritems(paramd): |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
278 if not k in ['PrecursorIonModeDefault', 'nm_mass_diff_default', 'additional_details']: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
279 cmd += " {}={}".format(str(k), str(v)) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
280 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
281 # =============== Run metfrag ============================================== |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
282 # Filter before process with a minimum number of MS/MS peaks |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
283 if plinesread >= float(args.minMSMSpeaks): |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
284 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
285 if int(args.cores_top_level) == 1: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
286 os.system(cmd) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
287 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
288 return paramd, cmd |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
289 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
290 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
291 def work(cmds): |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
292 return [os.system(cmd) for cmd in cmds] |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
293 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
294 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
295 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
296 ###################################################################### |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
297 # Parse MSP file and run metfrag CLI |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
298 ###################################################################### |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
299 # keep list of commands if performing in CLI in parallel |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
300 cmds = [] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
301 # keep a dictionary of all params |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
302 paramds = {} |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
303 # keep count of spectra (for uid) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
304 spectrac = 0 |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
305 # this dictionary will store the meta data results form the MSp file |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
306 meta_info = {} |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
307 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
308 with open(args.input_pth, "r") as infile: |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
309 # number of lines for the peaks |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
310 pnumlines = 0 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
311 # number of lines read for the peaks |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
312 plinesread = 0 |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
313 for line in infile: |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
314 line = line.strip() |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
315 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
316 if pnumlines == 0: |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
317 # =============== Extract metadata from MSP ======================== |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
318 meta_info = parse_meta(meta_regex, meta_info) |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
319 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
320 if ('massbank' in meta_info and 'cols' in meta_info) or ('msp' in meta_info and 'num_peaks' in meta_info): |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
321 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
322 pnumlines = int(meta_info['num_peaks']) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
323 plinesread = 0 |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
324 peaklist = [] |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
325 |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
326 elif plinesread < pnumlines: |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
327 # =============== Extract peaks from MSP ========================== |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
328 line = tuple(line.split()) # .split() will split on any empty space (i.e. tab and space) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
329 # Keep only m/z and intensity, not relative intensity |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
330 save_line = tuple(line[0].split() + line[1].split()) |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
331 plinesread += 1 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
332 peaklist.append(save_line) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
333 |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
334 elif plinesread and plinesread == pnumlines: |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
335 # =============== Get sample name and additional details for output ======= |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
336 spectrac += 1 |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
337 paramd, cmd = run_metfrag(meta_info, peaklist, args, wd, spectrac, adduct_types) |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
338 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
339 paramds[paramd["SampleName"]] = paramd |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
340 cmds.append(cmd) |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
341 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
342 meta_info = {} |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
343 pnumlines = 0 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
344 plinesread = 0 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
345 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
346 # end of file. Check if there is a MSP spectra to run metfrag on still |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
347 if plinesread and plinesread == pnumlines: |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
348 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
349 paramd, cmd = run_metfrag(meta_info, peaklist, args, wd, spectrac+1, adduct_types) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
350 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
351 paramds[paramd["SampleName"]] = paramd |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
352 cmds.append(cmd) |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
353 |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
354 |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
355 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
356 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
357 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
358 # Perform multiprocessing on command line call level |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
359 if int(args.cores_top_level) > 1: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
360 cmds_chunks = [cmds[x:x + int(args.chunks)] for x in list(range(0, len(cmds), int(args.chunks)))] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
361 pool = multiprocessing.Pool(processes=int(args.cores_top_level)) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
362 pool.map(work, cmds_chunks) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
363 pool.close() |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
364 pool.join() |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
365 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
366 ###################################################################### |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
367 # Concatenate and filter the output |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
368 ###################################################################### |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
369 # outputs might have different headers. Need to get a list of all the headers before we start merging the files |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
370 # outfiles = [os.path.join(wd, f) for f in glob.glob(os.path.join(wd, "*_metfrag_result.csv"))] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
371 outfiles = glob.glob(os.path.join(wd, "*_metfrag_result.csv")) |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
372 |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
373 headers = [] |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
374 c = 0 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
375 for fn in outfiles: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
376 with open(fn, 'r') as infile: |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
377 reader = csv.reader(infile) |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
378 if sys.version_info >= (3, 0): |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
379 headers.extend(next(reader)) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
380 else: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
381 headers.extend(reader.next()) |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
382 # check if file has any data rows |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
383 for i, row in enumerate(reader): |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
384 c += 1 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
385 if i == 1: |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
386 break |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
387 |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
388 # if no data rows (e.g. matches) then do not save an output and leave the program |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
389 if c == 0: |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
390 sys.exit() |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
391 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
392 additional_detail_headers = ['sample_name'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
393 for k, paramd in six.iteritems(paramds): |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
394 additional_detail_headers = list(set(additional_detail_headers + list(paramd['additional_details'].keys()))) |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
395 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
396 headers = additional_detail_headers + sorted(list(set(headers))) |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
397 |
|
2
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
398 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
399 # Sort files nicely |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
400 outfiles.sort(key = lambda s: int(re.match('^.*/(\d+)_metfrag_result.csv', s).group(1))) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
401 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
402 print(outfiles) |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
403 |
|
d040e27b6225
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents:
1
diff
changeset
|
404 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
405 # merge outputs |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
406 with open(args.result_pth, 'a') as merged_outfile: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
407 dwriter = csv.DictWriter(merged_outfile, fieldnames=headers, delimiter='\t', quotechar='"', |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
408 quoting=csv.QUOTE_NONNUMERIC,) |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
409 dwriter.writeheader() |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
410 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
411 for fn in outfiles: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
412 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
413 with open(fn) as infile: |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
414 reader = csv.DictReader(infile, delimiter=',', quotechar='"') |
|
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
415 for line in reader: |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
416 bewrite = True |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
417 for key, value in line.items(): |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
418 # Filter when no MS/MS peak matched |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
419 if key == "ExplPeaks": |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
420 if float(args.pctexplpeak_thrshld) > 0 and "NA" in value: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
421 bewrite = False |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
422 # Filter with a score threshold |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
423 elif key == "Score": |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
424 if float(value) <= float(args.score_thrshld): |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
425 bewrite = False |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
426 elif key == "NoExplPeaks": |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
427 nbfindpeak = float(value) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
428 elif key == "NumberPeaksUsed": |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
429 totpeaks = float(value) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
430 # Filter with a relative number of peak matched |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
431 try: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
432 pctexplpeak = nbfindpeak / totpeaks * 100 |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
433 except ZeroDivisionError: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
434 bewrite = False |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
435 else: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
436 if pctexplpeak < float(args.pctexplpeak_thrshld): |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
437 bewrite = False |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
438 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
439 # Write the line if it pass all filters |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
440 if bewrite: |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
441 bfn = os.path.basename(fn) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
442 bfn = bfn.replace(".csv", "") |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
443 ad = paramds[bfn]['additional_details'] |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
444 line.update(ad) |
|
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
445 line['sample_name'] = paramds[bfn]['SampleName'] |
|
0
75c805123b45
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff
changeset
|
446 |
|
1
c1b168770b68
planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents:
0
diff
changeset
|
447 dwriter.writerow(line) |
