Mercurial > repos > iuc > genetrack

annotate genetrack.py @ 2:aaca27a5263b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author iuc
date Fri, 13 Jan 2017 10:45:36 -0500
parents 25cd59a002d9
children 41887967ef14
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
1 """
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
2 genetrack.py
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
3
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
4 Input: either scidx or gff format of reads
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
5 Output: Called peaks in gff format
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
6 """
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
7 import optparse
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
8 import csv
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
9 import os
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
10 import genetrack_util
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
11
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
12 CHUNK_SIZE = 10000000
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
13
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
14
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
15 if __name__ == '__main__':
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
16 parser = optparse.OptionParser()
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
17 parser.add_option('-t', '--input_format', dest='input_format', type='string', help='Input format')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
18 parser.add_option('-i', '--input', dest='inputs', type='string', action='append', nargs=2, help='Input datasets')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
19 parser.add_option('-s', '--sigma', dest='sigma', type='int', default=5, help='Sigma.')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
20 parser.add_option('-e', '--exclusion', dest='exclusion', type='int', default=20, help='Exclusion zone.')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
21 parser.add_option('-u', '--up_width', dest='up_width', type='int', default=10, help='Upstream width of called peaks.')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
22 parser.add_option('-d', '--down_width', dest='down_width', type='int', default=10, help='Downstream width of called peaks.')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
23 parser.add_option('-f', '--filter', dest='filter', type='int', default=1, help='Absolute read filter.')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
24 options, args = parser.parse_args()
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
25
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
26 os.mkdir('output')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
27 for (dataset_path, hid) in options.inputs:
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
28 if options.input_format == 'gff':
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
29 # Make sure the reads for each chromosome are sorted by index.
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
30 input_path = genetrack_util.sort_chromosome_reads_by_index(dataset_path)
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
31 else:
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
32 # We're processing scidx data.
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
33 input_path = dataset_path
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
34 output_name = 's%se%su%sd%sF%s_on_data_%s' % (options.sigma,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
35 options.exclusion,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
36 options.up_width,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
37 options.down_width,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
38 options.filter,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
39 hid)
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
40 output_path = os.path.join('output', output_name)
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
41 reader = csv.reader(open(input_path, 'rU'), delimiter='\t')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
42 writer = csv.writer(open(output_path, 'wt'), delimiter='\t')
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
43 width = options.sigma * 5
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
44 manager = genetrack_util.ChromosomeManager(reader)
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
45 while not manager.done:
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
46 cname = manager.chromosome_name()
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
47 # Should we process this chromosome?
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
48 data = manager.load_chromosome()
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
49 if not data:
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
50 continue
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
51 keys = genetrack_util.make_keys(data)
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
52 lo, hi = genetrack_util.get_range(data)
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
53 for chunk in genetrack_util.get_chunks(lo, hi, size=CHUNK_SIZE, overlap=width):
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
54 (slice_start, slice_end), process_bounds = chunk
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
55 window = genetrack_util.get_window(data, slice_start, slice_end, keys)
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
56 genetrack_util.process_chromosome(cname,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
57 window,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
58 writer,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
59 process_bounds,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
60 width,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
61 options.sigma,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
62 options.up_width,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
63 options.down_width,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
64 options.exclusion,
25cd59a002d9 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/genetrack commit e96df94dba60050fa28aaf55b5bb095717a5f260
iuc
parents:
diff changeset
65 options.filter)