Mercurial > repos > devteam > intersect
annotate gops_intersect.py @ 4:8ddabc73af92 draft
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit 4e2052686dfe8003f867449e0affff96398b2a62
| author | devteam |
|---|---|
| date | Thu, 11 Feb 2016 12:11:25 -0500 |
| parents | 5f72be09cfd3 |
| children | 33b3f3688db4 |
| rev | line source |
|---|---|
| 0 | 1 #!/usr/bin/env python |
| 2 """ | |
| 3 Find regions of first interval file that overlap regions in a second interval file. | |
| 4 Interval files can either be BED or GFF format. | |
| 5 | |
| 6 usage: %prog interval_file_1 interval_file_2 out_file | |
| 7 -1, --cols1=N,N,N,N: Columns for start, end, strand in first file | |
| 8 -2, --cols2=N,N,N,N: Columns for start, end, strand in second file | |
| 9 -m, --mincols=N: Require this much overlap (default 1bp) | |
| 10 -p, --pieces: just print pieces of second set (after padding) | |
| 11 -G, --gff1: input 1 is GFF format, meaning start and end coordinates are 1-based, closed interval | |
| 12 -H, --gff2: input 2 is GFF format, meaning start and end coordinates are 1-based, closed interval | |
| 13 """ | |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
14 import fileinput |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
15 import sys |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
16 from bx.intervals.io import GenomicInterval, NiceReaderWrapper |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
17 from bx.intervals.operations.intersect import intersect |
| 0 | 18 from bx.cookbook import doc_optparse |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
19 from bx.tabular.io import ParseError |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
20 from galaxy.tools.util.galaxyops import fail, parse_cols_arg, skipped |
| 0 | 21 from utils.gff_util import GFFFeature, GFFReaderWrapper, convert_bed_coords_to_gff |
| 22 | |
| 23 assert sys.version_info[:2] >= ( 2, 4 ) | |
| 24 | |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
25 |
| 0 | 26 def main(): |
| 27 mincols = 1 | |
| 28 | |
| 29 options, args = doc_optparse.parse( __doc__ ) | |
| 30 try: | |
| 31 chr_col_1, start_col_1, end_col_1, strand_col_1 = parse_cols_arg( options.cols1 ) | |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
32 chr_col_2, start_col_2, end_col_2, strand_col_2 = parse_cols_arg( options.cols2 ) |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
33 if options.mincols: |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
34 mincols = int( options.mincols ) |
| 0 | 35 pieces = bool( options.pieces ) |
| 36 in1_gff_format = bool( options.gff1 ) | |
| 37 in2_gff_format = bool( options.gff2 ) | |
| 38 in_fname, in2_fname, out_fname = args | |
| 39 except: | |
| 40 doc_optparse.exception() | |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
41 |
| 0 | 42 # Set readers to handle either GFF or default format. |
| 43 if in1_gff_format: | |
| 44 in1_reader_wrapper = GFFReaderWrapper | |
| 45 else: | |
| 46 in1_reader_wrapper = NiceReaderWrapper | |
| 47 if in2_gff_format: | |
| 48 in2_reader_wrapper = GFFReaderWrapper | |
| 49 else: | |
| 50 in2_reader_wrapper = NiceReaderWrapper | |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
51 |
| 0 | 52 g1 = in1_reader_wrapper( fileinput.FileInput( in_fname ), |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
53 chrom_col=chr_col_1, |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
54 start_col=start_col_1, |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
55 end_col=end_col_1, |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
56 strand_col=strand_col_1, |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
57 fix_strand=True ) |
| 0 | 58 if in1_gff_format: |
| 59 # Intersect requires coordinates in BED format. | |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
60 g1.convert_to_bed_coord = True |
| 0 | 61 g2 = in2_reader_wrapper( fileinput.FileInput( in2_fname ), |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
62 chrom_col=chr_col_2, |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
63 start_col=start_col_2, |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
64 end_col=end_col_2, |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
65 strand_col=strand_col_2, |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
66 fix_strand=True ) |
| 0 | 67 if in2_gff_format: |
| 68 # Intersect requires coordinates in BED format. | |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
69 g2.convert_to_bed_coord = True |
|
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
70 |
| 0 | 71 out_file = open( out_fname, "w" ) |
| 72 try: | |
|
3
5f72be09cfd3
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents:
0
diff
changeset
|
73 for feature in intersect( [g1, g2], pieces=pieces, mincols=mincols ): |
| 0 | 74 if isinstance( feature, GFFFeature ): |
| 75 # Convert back to GFF coordinates since reader converted automatically. | |
| 76 convert_bed_coords_to_gff( feature ) | |
| 77 for interval in feature.intervals: | |
| 78 out_file.write( "%s\n" % "\t".join( interval.fields ) ) | |
| 79 elif isinstance( feature, GenomicInterval ): | |
| 80 out_file.write( "%s\n" % "\t".join( feature.fields ) ) | |
| 81 else: | |
| 82 out_file.write( "%s\n" % feature ) | |
| 83 except ParseError, e: | |
| 84 out_file.close() | |
| 85 fail( "Invalid file format: %s" % str( e ) ) | |
| 86 | |
| 87 out_file.close() | |
| 88 | |
| 89 if g1.skipped > 0: | |
| 90 print skipped( g1, filedesc=" of 1st dataset" ) | |
| 91 if g2.skipped > 0: | |
| 92 print skipped( g2, filedesc=" of 2nd dataset" ) | |
| 93 | |
| 94 if __name__ == "__main__": | |
| 95 main() |
