Mercurial > repos > devteam > merge
comparison gops_merge.py @ 0:10ac6097acdb
Imported from capsule None
| author | devteam |
|---|---|
| date | Tue, 01 Apr 2014 09:13:25 -0400 |
| parents | |
| children | b9c97d3233bb |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:10ac6097acdb |
|---|---|
| 1 #!/usr/bin/env python | |
| 2 """ | |
| 3 Merge overlaping regions. | |
| 4 | |
| 5 usage: %prog in_file out_file | |
| 6 -1, --cols1=N,N,N,N: Columns for start, end, strand in first file | |
| 7 -m, --mincols=N: Require this much overlap (default 1bp) | |
| 8 -3, --threecol: Output 3 column bed | |
| 9 """ | |
| 10 import sys, traceback, fileinput | |
| 11 from warnings import warn | |
| 12 from bx.intervals import * | |
| 13 from bx.intervals.io import * | |
| 14 from bx.intervals.operations.merge import * | |
| 15 from bx.cookbook import doc_optparse | |
| 16 from galaxy.tools.util.galaxyops import * | |
| 17 | |
| 18 assert sys.version_info[:2] >= ( 2, 4 ) | |
| 19 | |
| 20 def main(): | |
| 21 mincols = 1 | |
| 22 upstream_pad = 0 | |
| 23 downstream_pad = 0 | |
| 24 | |
| 25 options, args = doc_optparse.parse( __doc__ ) | |
| 26 try: | |
| 27 chr_col_1, start_col_1, end_col_1, strand_col_1 = parse_cols_arg( options.cols1 ) | |
| 28 if options.mincols: mincols = int( options.mincols ) | |
| 29 in_fname, out_fname = args | |
| 30 except: | |
| 31 doc_optparse.exception() | |
| 32 | |
| 33 g1 = NiceReaderWrapper( fileinput.FileInput( in_fname ), | |
| 34 chrom_col=chr_col_1, | |
| 35 start_col=start_col_1, | |
| 36 end_col=end_col_1, | |
| 37 strand_col = strand_col_1, | |
| 38 fix_strand=True ) | |
| 39 | |
| 40 out_file = open( out_fname, "w" ) | |
| 41 | |
| 42 try: | |
| 43 for line in merge(g1,mincols=mincols): | |
| 44 if options.threecol: | |
| 45 if type( line ) is GenomicInterval: | |
| 46 out_file.write( "%s\t%s\t%s\n" % ( line.chrom, str( line.startCol ), str( line.endCol ) ) ) | |
| 47 elif type( line ) is list: | |
| 48 out_file.write( "%s\t%s\t%s\n" % ( line[chr_col_1], str( line[start_col_1] ), str( line[end_col_1] ) ) ) | |
| 49 else: | |
| 50 out_file.write( "%s\n" % line ) | |
| 51 else: | |
| 52 if type( line ) is GenomicInterval: | |
| 53 out_file.write( "%s\n" % "\t".join( line.fields ) ) | |
| 54 elif type( line ) is list: | |
| 55 out_file.write( "%s\n" % "\t".join( line ) ) | |
| 56 else: | |
| 57 out_file.write( "%s\n" % line ) | |
| 58 except ParseError, exc: | |
| 59 out_file.close() | |
| 60 fail( "Invalid file format: %s" % str( exc ) ) | |
| 61 | |
| 62 out_file.close() | |
| 63 | |
| 64 if g1.skipped > 0: | |
| 65 print skipped( g1, filedesc=" of 1st dataset" ) | |
| 66 | |
| 67 if __name__ == "__main__": | |
| 68 main() |