annotate sRbowtieParser.py @ 17:7b5e0fba8e74 draft default tip

planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/msp_sr_bowtie_parser commit 30a73e24eabb05f9d46150fd03471ff853579ad2-dirty
author mvdbeek
date Mon, 18 Apr 2016 09:52:54 -0400
parents 70193ce0540e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
1 #!/usr/bin/python
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
2 # python parser module to analyse sRbowtie alignments
2
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
3 # version 1.0.2 - argparse implementation
0
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
4 # Usage sRbowtieParser.py <1:index source> <2:extraction directive> <3:outputL> <4:polarity> <5:6:7 filePath:FileExt:FileLabel> <.. ad lib>
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
5
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
6 import sys, argparse
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
7 from smRtools import *
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
8
2
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
9 def masterListGenerator(data_source):
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
10 for filePath, FileExt, FileLabel in data_source:
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
11 yield HandleSmRNAwindows (filePath, FileExt, IndexSource, genomeRefFormat)
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
12
0
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
13 def Parser():
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
14 the_parser = argparse.ArgumentParser()
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
15 the_parser.add_argument('--IndexSource', action="store", type=str, help="Path to the index source")
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
16 the_parser.add_argument('--ExtractDirective', action="store", type=str, choices=["fastaSource", "bowtieIndex"], help="Extract info from fasta or bowtie index")
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
17 the_parser.add_argument('--output', action="store", type=str, help="path to the output")
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
18 the_parser.add_argument('--polarity', choices=["forward", "reverse", "both"], help="forward, reverse or both forward an reverse reads are counted")
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
19 the_parser.add_argument('--alignmentSource',nargs='+', help="paths to alignments files")
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
20 the_parser.add_argument('--alignmentFormat',nargs='+', help="Format of the bowtie alignment (tabular, sam or bam)")
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
21 the_parser.add_argument('--alignmentLabel',nargs='+', help="Label of the alignment")
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
22 args = the_parser.parse_args()
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
23 return args
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
24
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
25 args = Parser()
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
26
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
27 IndexSource = args.IndexSource
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
28 genomeRefFormat = args.ExtractDirective
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
29 Output = args.output
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
30 Polarity = args.polarity
2
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
31 header = ["gene"]
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
32
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
33
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
34 FileLabelList=[label for label in args.alignmentLabel]
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
35 header.extend(FileLabelList)
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
36 assert (len(FileLabelList)==len(set(FileLabelList))),"You have supplied a non-unique label. Please make sure that your input files have unique names"
0
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
37
2
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
38 data_source=zip (args.alignmentSource, args.alignmentFormat, args.alignmentLabel)
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
39 master_generator=masterListGenerator(data_source)
0
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
40
2
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
41 for i,window in enumerate(master_generator):
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
42 window=window
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
43 if i==0:
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
44 gene_count_dict={gene:[str(item.readcount(polarity=Polarity))] for gene,item in window.instanceDict.items()}
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
45 else:
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
46 [gene_count_dict[gene].append(str(item.readcount(polarity=Polarity))) for gene,item in window.instanceDict.items()]
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
47
0
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
48
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
49 F = open (args.output, "w")
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
50 # print >>F, args
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
51 print >> F, "\t".join(header)
2
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
52
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
53 for item in sorted(gene_count_dict.keys()):
0
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
54 line=[item]
2
70193ce0540e planemo upload for repository https://bitbucket.org/drosofff/gedtools/
mvdbeek
parents: 0
diff changeset
55 line.extend(gene_count_dict[item])
0
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
56 print >> F, "\t".join(line )
3a510730e3fc Imported from capsule None
drosofff
parents:
diff changeset
57 F.close()