annotate metfrag.py @ 5:c53ed894d736 draft

planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 826659b9105b635cf240474428122f5a269f5c88
author tomnl
date Fri, 02 Aug 2019 11:38:06 -0400
parents eb581a101672
children 0b3816a7a14b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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')
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
53 parser.add_argument('--skip_invalid_adducts', action='store_true')
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
54
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
55 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
56 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
57 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
58
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
59
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
60 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
61 print(args)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
62
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
63 if os.stat(args.input_pth).st_size == 0:
5
c53ed894d736 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 826659b9105b635cf240474428122f5a269f5c88
tomnl
parents: 4
diff changeset
64 print('Input file empty')
c53ed894d736 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 826659b9105b635cf240474428122f5a269f5c88
tomnl
parents: 4
diff changeset
65 exit()
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
66
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
67
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
68 # Create temporary working directory
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
69 if args.temp_dir:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
70 wd = args.temp_dir
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
71 else:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
72 wd = tempfile.mkdtemp()
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 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
75 shutil.rmtree(wd)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
76 os.makedirs(wd)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
77 else:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
78 os.makedirs(wd)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
79
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
80 ######################################################################
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
81 # 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
82 ######################################################################
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
83 regex_msp = {}
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
84 regex_msp['name'] = ['^Name(?:=|:)(.*)$']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
85 regex_msp['polarity'] = ['^ion.*mode(?:=|:)(.*)$', '^ionization.*mode(?:=|:)(.*)$', '^polarity(?:=|:)(.*)$']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
86 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
87 regex_msp['precursor_type'] = ['^precursor.*type(?:=|:)(.*)$', '^adduct(?:=|:)(.*)$', '^ADDUCTIONNAME(?:=|:)(.*)$']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
88 regex_msp['num_peaks'] = ['^Num.*Peaks(?:=|:)\s*(\d*)$']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
89 regex_msp['msp'] = ['^Name(?:=|:)(.*)$'] # Flag for standard MSP format
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
90
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
91 regex_massbank = {}
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
92 regex_massbank['name'] = ['^RECORD_TITLE:(.*)$']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
93 regex_massbank['polarity'] = ['^AC\$MASS_SPECTROMETRY:\s+ION_MODE\s+(.*)$']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
94 regex_massbank['precursor_mz'] = ['^MS\$FOCUSED_ION:\s+PRECURSOR_M/Z\s+(\d*[.,]?\d*)$']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
95 regex_massbank['precursor_type'] = ['^MS\$FOCUSED_ION:\s+PRECURSOR_TYPE\s+(.*)$']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
96 regex_massbank['num_peaks'] = ['^PK\$NUM_PEAK:\s+(\d*)']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
97 regex_massbank['cols'] = ['^PK\$PEAK:\s+(.*)']
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
98 regex_massbank['massbank'] = ['^RECORD_TITLE:(.*)$'] # Flag for massbank format
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
99
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
100 if args.schema == 'msp':
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
101 meta_regex = regex_msp
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
102 elif args.schema == 'massbank':
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
103 meta_regex = regex_massbank
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
104 elif args.schema == 'auto':
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
105 # 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
106 # the name parameter
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
107 meta_regex = {}
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
108 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
109 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
110 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
111 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
112 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
113 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
114 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
115
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
116 print(meta_regex)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
117
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
118 adduct_types = {
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
119 '[M+H]+': 1.007276,
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
120 '[M+NH4]+': 18.034374,
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
121 '[M+Na]+': 22.989218,
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
122 '[M+K]+': 38.963158,
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
123 '[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
124 '[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
125 '[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
126 '[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
127 '[M-H]-': -1.007276,
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
128 '[M+Cl]-': 34.969402,
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
129 '[M+HCOO]-': 44.99819,
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
130 '[M-H+HCOOH]-': 44.99819, # same as above but different style of writing adduct
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
131 '[M+CH3COO]-': 59.01385,
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
132 '[M-H+CH3COOH]-': 59.01385 # same as above but different style of writing adduct
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
133 }
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
134
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
135 # 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
136 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
137 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
138 for reg in regexes:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
139 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
140 if m:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
141 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
142 return meta_info
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
143
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
144
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
145
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
146 ######################################################################
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
147 # Setup parameter dictionary
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
148 ######################################################################
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
149 def init_paramd(args):
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
150 paramd = defaultdict()
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["MetFragDatabaseType"] = args.MetFragDatabaseType
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
153
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
154 if args.MetFragDatabaseType == "LocalCSV":
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
155 paramd["LocalDatabasePath"] = args.LocalDatabasePath
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
156 elif args.MetFragDatabaseType == "MetChem":
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
157 paramd["LocalMetChemDatabase"] = "metchem"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
158 paramd["LocalMetChemDatabasePortNumber"] = 5432
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
159 paramd["LocalMetChemDatabaseServerIp"] = args.LocalMetChemDatabaseServerIp
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
160 paramd["LocalMetChemDatabaseUser"] = "metchemro"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
161 paramd["LocalMetChemDatabasePassword"] = "metchemro"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
162
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
163 paramd["FragmentPeakMatchAbsoluteMassDeviation"] = args.FragmentPeakMatchAbsoluteMassDeviation
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
164 paramd["FragmentPeakMatchRelativeMassDeviation"] = args.FragmentPeakMatchRelativeMassDeviation
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
165 paramd["DatabaseSearchRelativeMassDeviation"] = args.DatabaseSearchRelativeMassDeviation
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
166 paramd["SampleName"] = ''
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
167 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
168
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
169 if args.polarity == "pos":
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
170 paramd["IsPositiveIonMode"] = True
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
171 paramd["PrecursorIonModeDefault"] = "1"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
172 paramd["PrecursorIonMode"] = "1"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
173 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
174 else:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
175 paramd["IsPositiveIonMode"] = False
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
176 paramd["PrecursorIonModeDefault"] = "-1"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
177 paramd["PrecursorIonMode"] = "-1"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
178 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
179
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
180 paramd["MetFragCandidateWriter"] = "CSV"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
181 paramd["NumberThreads"] = args.NumberThreads
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
182
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
183 if args.ScoreSuspectLists:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
184 paramd["ScoreSuspectLists"] = args.ScoreSuspectLists
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
185
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
186 paramd["MetFragScoreTypes"] = args.MetFragScoreTypes
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
187 paramd["MetFragScoreWeights"] = args.MetFragScoreWeights
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
188
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
189 dct_filter = defaultdict()
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
190 filterh = []
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
191
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
192 if args.UnconnectedCompoundFilter:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
193 filterh.append('UnconnectedCompoundFilter')
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.IsotopeFilter:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
196 filterh.append('IsotopeFilter')
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
197
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
198 if args.FilterMinimumElements:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
199 filterh.append('MinimumElementsFilter')
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
200 dct_filter['FilterMinimumElements'] = args.FilterMinimumElements
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
201
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
202 if args.FilterMaximumElements:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
203 filterh.append('MaximumElementsFilter')
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
204 dct_filter['FilterMaximumElements'] = args.FilterMaximumElements
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
205
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
206 if args.FilterSmartsInclusionList:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
207 filterh.append('SmartsSubstructureInclusionFilter')
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
208 dct_filter['FilterSmartsInclusionList'] = args.FilterSmartsInclusionList
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
209
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
210 if args.FilterSmartsExclusionList:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
211 filterh.append('SmartsSubstructureExclusionFilter')
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
212 dct_filter['FilterSmartsExclusionList'] = args.FilterSmartsExclusionList
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
213
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
214 # 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
215 # 'FilterIncludedElements'
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
216 if args.FilterIncludedExclusiveElements:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
217 filterh.append('ElementInclusionExclusiveFilter')
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
218 dct_filter['FilterIncludedElements'] = args.FilterIncludedExclusiveElements
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
219
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
220 if args.FilterIncludedElements:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
221 filterh.append('ElementInclusionFilter')
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
222 dct_filter['FilterIncludedElements'] = args.FilterIncludedElements
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 if args.FilterExcludedElements:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
225 filterh.append('ElementExclusionFilter')
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
226 dct_filter['FilterExcludedElements'] = args.FilterExcludedElements
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 if filterh:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
229 fcmds = ','.join(filterh) + ' '
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
230 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
231 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
232
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
233 paramd["MetFragPreProcessingCandidateFilter"] = fcmds
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
234
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
235 return paramd
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
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
238 ######################################################################
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
239 # 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
240 ######################################################################
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
241 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
242 # 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
243 # 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
244 # 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
245 # just the spectra index (spectrac)]
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
246 # 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
247
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
248 paramd = init_paramd(args)
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
249 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
250 # 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
251 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
252 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
253 # 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
254 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
255 meta_info['name'].split("|")}
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
256 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
257 # 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
258 paramd['additional_details'] = meta_info
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
259 else:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
260 # 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
261 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
262
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
263 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
264
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
265 # =============== Output peaks to txt file ==============================
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
266 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
267
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
268 # write spec file
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
269 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
270 for p in peaklist:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
271 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
272
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
273 # =============== 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
274 # 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
275 if 'precursor_type' in meta_info and meta_info['precursor_type'] in adduct_types:
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
276 nm = float(meta_info['precursor_mz']) - adduct_types[meta_info['precursor_type']]
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
277 paramd["PrecursorIonMode"] = int(round(adduct_types[meta_info['precursor_type']], 0))
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
278 elif not args.skip_invalid_adducts:
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
279 paramd["PrecursorIonMode"] = paramd['PrecursorIonModeDefault']
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
280 nm = float(meta_info['precursor_mz']) - paramd['nm_mass_diff_default']
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
281 else:
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
282 print('Skipping {}'.format(paramd["SampleName"]))
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
283 return '', ''
2
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 paramd["NeutralPrecursorMass"] = nm
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
286
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
287 # =============== Create CLI cmd for metfrag ===============================
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
288 cmd = "metfrag"
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
289 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
290 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
291 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
292
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
293 # =============== Run metfrag ==============================================
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
294 print(cmd)
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
295 # 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
296 if plinesread >= float(args.minMSMSpeaks):
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
297
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
298 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
299 os.system(cmd)
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
300
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
301 return paramd, cmd
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
302
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
303
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
304 def work(cmds):
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
305 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
306
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
307
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
308
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
309 ######################################################################
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
310 # 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
311 ######################################################################
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
312 # 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
313 cmds = []
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
314 # 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
315 paramds = {}
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
316 # 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
317 spectrac = 0
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
318 # 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
319 meta_info = {}
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
320
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
321 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
322 # 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
323 pnumlines = 0
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
324 # 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
325 plinesread = 0
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
326 for line in infile:
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
327 line = line.strip()
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
328
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
329 if pnumlines == 0:
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
330 # =============== Extract metadata from MSP ========================
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
331 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
332
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
333 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
334
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
335 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
336 plinesread = 0
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
337 peaklist = []
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
338
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
339 elif plinesread < pnumlines:
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
340 # =============== Extract peaks from MSP ==========================
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
341 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
342 # 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
343 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
344 plinesread += 1
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
345 peaklist.append(save_line)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
346
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
347 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
348 # =============== 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
349 spectrac += 1
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
350 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
351
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
352 if paramd:
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
353 paramds[paramd["SampleName"]] = paramd
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
354 cmds.append(cmd)
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
355
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
356 meta_info = {}
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
357 pnumlines = 0
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
358 plinesread = 0
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
359
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
360 # 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
361 if plinesread and plinesread == pnumlines:
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
362
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
363 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
364
4
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
365 if paramd:
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
366 paramds[paramd["SampleName"]] = paramd
eb581a101672 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 28da8cdf5f9cc45575b614a329da5790803e59c4
tomnl
parents: 3
diff changeset
367 cmds.append(cmd)
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
368
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
369
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
370
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
371
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
372 # 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
373 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
374 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
375 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
376 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
377 pool.close()
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
378 pool.join()
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
379
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
380 ######################################################################
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
381 # 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
382 ######################################################################
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
383 # 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
384 # 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
385 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
386
5
c53ed894d736 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 826659b9105b635cf240474428122f5a269f5c88
tomnl
parents: 4
diff changeset
387 if len(outfiles) == 0:
c53ed894d736 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 826659b9105b635cf240474428122f5a269f5c88
tomnl
parents: 4
diff changeset
388 print('No results')
c53ed894d736 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 826659b9105b635cf240474428122f5a269f5c88
tomnl
parents: 4
diff changeset
389 sys.exit()
c53ed894d736 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 826659b9105b635cf240474428122f5a269f5c88
tomnl
parents: 4
diff changeset
390
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
391 headers = []
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
392 c = 0
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
393 for fn in outfiles:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
394 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
395 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
396 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
397 headers.extend(next(reader))
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
398 else:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
399 headers.extend(reader.next())
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
400 # 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
401 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
402 c += 1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
403 if i == 1:
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
404 break
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
405
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
406 # 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
407 if c == 0:
5
c53ed894d736 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 826659b9105b635cf240474428122f5a269f5c88
tomnl
parents: 4
diff changeset
408 print('No results')
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
409 sys.exit()
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 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
412 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
413 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
414
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
415 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
416
2
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
417
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
418 # Sort files nicely
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
419 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
420
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
421 print(outfiles)
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
422
d040e27b6225 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 59631850110e60900d03f642c445cc35a36414b5
tomnl
parents: 1
diff changeset
423
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
424 # merge outputs
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
425 with open(args.result_pth, 'a') as merged_outfile:
3
5ee936e570a7 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b5d5fdc81a1756000dc25d4aa672fec1ece732a
tomnl
parents: 2
diff changeset
426 dwriter = csv.DictWriter(merged_outfile, fieldnames=headers, delimiter='\t', quotechar='"')
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
427 dwriter.writeheader()
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
428
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
429 for fn in outfiles:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
430
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
431 with open(fn) as infile:
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
432 reader = csv.DictReader(infile, delimiter=',', quotechar='"')
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
433 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
434 bewrite = True
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
435 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
436 # 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
437 if key == "ExplPeaks":
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
438 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
439 bewrite = False
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
440 # 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
441 elif key == "Score":
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
442 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
443 bewrite = False
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
444 elif key == "NoExplPeaks":
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
445 nbfindpeak = float(value)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
446 elif key == "NumberPeaksUsed":
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
447 totpeaks = float(value)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
448 # 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
449 try:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
450 pctexplpeak = nbfindpeak / totpeaks * 100
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
451 except ZeroDivisionError:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
452 bewrite = False
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
453 else:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
454 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
455 bewrite = False
0
75c805123b45 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit b5f4005c77b275b9b1b4349b08476afc783a8e14
tomnl
parents:
diff changeset
456
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
457 # 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
458 if bewrite:
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
459 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
460 bfn = bfn.replace(".csv", "")
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
461 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
462 line.update(ad)
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
463 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
464
1
c1b168770b68 planemo upload for repository https://github.com/computational-metabolomics/metfrag-galaxy commit 5b384bd55e5bb4b0dc1daebd2ef5d3ee0e379b2e-dirty
tomnl
parents: 0
diff changeset
465 dwriter.writerow(line)