annotate link_scan_datasets.py @ 5:274fdc50169b draft default tip

planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
author jjohnson
date Wed, 25 Jul 2018 15:05:34 -0400
parents 7f02fc51bddf
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
1 #!/usr/bin/env python
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
2
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
3 from __future__ import print_function
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
4
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
5 import argparse
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
6 import os
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
7 import os.path
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
8 import xml.sax
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
9 from difflib import SequenceMatcher
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
10
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
11
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
12 def __main__():
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
13 parser = argparse.ArgumentParser(
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
14 description='link spectrum datasets to the name used' +
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
15 ' in the identification dataset')
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
16 parser.add_argument(
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
17 'ident_files', nargs='+',
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
18 help='Pepxml or mzIdentML')
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
19 parser.add_argument(
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
20 '-n', '--scan_name', default=[], action='append',
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
21 help='Name for scan file')
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
22 parser.add_argument(
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
23 '-f', '--scan_file', default=[], action='append',
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
24 help='Path for scan file')
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
25 args = parser.parse_args()
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
26
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
27 class MzidHandler(xml.sax.ContentHandler):
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
28
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
29 def __init__(self):
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
30 xml.sax.ContentHandler.__init__(self)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
31 self.spectraDataFiles = []
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
32 self.spectraDataNames = []
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
33 self.searchDatabaseFiles = []
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
34 self.searchDatabaseNames = []
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
35
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
36 def startElement(self, tag, attrs):
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
37 if tag == 'SpectraData':
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
38 id = attrs['id']
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
39 path = attrs['location']
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
40 filename = os.path.basename(path)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
41 name = attrs['name'] if 'name' in attrs else None
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
42 self.spectraDataFiles.append(filename)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
43 self.spectraDataNames.append(name if name else id)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
44 print ("SpectraData: %s %s" % (name if name else id, path))
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
45 if tag == 'SearchDatabase':
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
46 id = attrs['id']
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
47 path = attrs['location']
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
48 filename = os.path.basename(path)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
49 name = attrs['name'] if 'name' in attrs else None
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
50 self.searchDatabaseFiles.append(filename)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
51 self.searchDatabaseNames.append(name if name else id)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
52 print ("SearchDatabase: %s %s" % (name if name else id, path))
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
53
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
54 def endElement(self, name):
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
55 pass
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
56
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
57 def characters(self, data):
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
58 pass
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
59
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
60 class PepXmlHandler(xml.sax.ContentHandler):
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
61
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
62 def __init__(self):
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
63 xml.sax.ContentHandler.__init__(self)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
64 self.spectraDataFiles = []
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
65 self.spectraDataNames = []
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
66
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
67 def startElement(self, tag, attrs):
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
68 if tag == 'msms_run_summary':
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
69 basename = attrs['base_name']
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
70 name = os.path.basename(basename)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
71 ext = attrs['raw_data']
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
72 path = '%s%s' % (basename, ext)
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
73 filename = os.path.basename(path)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
74 self.spectraDataFiles.append(filename)
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
75 self.spectraDataNames.append(name)
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
76 print ("SpectraData: %s %s" % (name, path))
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
77
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
78 def endElement(self, name):
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
79 pass
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
80
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
81 def characters(self, data):
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
82 pass
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
83
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
84 parser = xml.sax.make_parser()
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
85 parser.setFeature(xml.sax.handler.feature_namespaces, 0)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
86 handler = PepXmlHandler()
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
87 parser.setContentHandler(handler)
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
88 for ident in args.ident_files:
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
89 parser.parse(ident)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
90
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
91 spectra_names = handler.spectraDataFiles
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
92
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
93 def best_match(name):
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
94 if name in spectra_names:
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
95 return name
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
96 try:
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
97 r = [SequenceMatcher(None, name, spectra_names[x]).ratio()
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
98 for x in range(len(spectra_names))]
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
99 return spectra_names[r.index(max(r))]
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
100 except Exception, e:
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
101 print ("best_match: %s %s" % (name, e))
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
102
5
274fdc50169b planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit e5b5b15b0a995c8588ff62f92fd0a2329fb7a6a5-dirty
jjohnson
parents: 3
diff changeset
103 for i, name in enumerate(args.scan_name):
3
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
104 path = args.scan_file[i] if len(args.scan_file) > i else ''
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
105 (root, ext) = os.path.splitext(name)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
106 print ("SpectraFile: %s %s" % (name, path))
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
107 iname = best_match(name)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
108 print ("IdentName: %s %s" % (name, iname))
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
109 if not os.path.exists(iname) and os.path.exists(path):
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
110 os.symlink(path, iname)
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
111 print ("%s -> %s" % (iname, path))
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
112
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
113
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
114 if __name__ == "__main__":
7f02fc51bddf planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/spectrast commit 379705f578f9a0465f497894c7d2b5f68b6a55e6-dirty
jjohnson
parents:
diff changeset
115 __main__()