Mercurial > repos > drosofff > msp_sr_bowtie_parser
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 |
| rev | line source |
|---|---|
| 0 | 1 #!/usr/bin/python |
| 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 | 4 # Usage sRbowtieParser.py <1:index source> <2:extraction directive> <3:outputL> <4:polarity> <5:6:7 filePath:FileExt:FileLabel> <.. ad lib> |
| 5 | |
| 6 import sys, argparse | |
| 7 from smRtools import * | |
| 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 | 13 def Parser(): |
| 14 the_parser = argparse.ArgumentParser() | |
| 15 the_parser.add_argument('--IndexSource', action="store", type=str, help="Path to the index source") | |
| 16 the_parser.add_argument('--ExtractDirective', action="store", type=str, choices=["fastaSource", "bowtieIndex"], help="Extract info from fasta or bowtie index") | |
| 17 the_parser.add_argument('--output', action="store", type=str, help="path to the output") | |
| 18 the_parser.add_argument('--polarity', choices=["forward", "reverse", "both"], help="forward, reverse or both forward an reverse reads are counted") | |
| 19 the_parser.add_argument('--alignmentSource',nargs='+', help="paths to alignments files") | |
| 20 the_parser.add_argument('--alignmentFormat',nargs='+', help="Format of the bowtie alignment (tabular, sam or bam)") | |
| 21 the_parser.add_argument('--alignmentLabel',nargs='+', help="Label of the alignment") | |
| 22 args = the_parser.parse_args() | |
| 23 return args | |
| 24 | |
| 25 args = Parser() | |
| 26 | |
| 27 IndexSource = args.IndexSource | |
| 28 genomeRefFormat = args.ExtractDirective | |
| 29 Output = args.output | |
| 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 | 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 | 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 | 48 |
| 49 F = open (args.output, "w") | |
| 50 # print >>F, args | |
| 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 | 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 | 56 print >> F, "\t".join(line ) |
| 57 F.close() |
