[2] | 1 | #!/usr/bin/env python |
---|
| 2 | |
---|
| 3 | """ |
---|
| 4 | Read a maf and output intervals for specified list of species. |
---|
| 5 | """ |
---|
| 6 | import sys, os |
---|
| 7 | from galaxy import eggs |
---|
| 8 | import pkg_resources; pkg_resources.require( "bx-python" ) |
---|
| 9 | from bx.align import maf |
---|
| 10 | from galaxy.tools.util import maf_utilities |
---|
| 11 | |
---|
| 12 | assert sys.version_info[:2] >= ( 2, 4 ) |
---|
| 13 | |
---|
| 14 | def __main__(): |
---|
| 15 | input_filename = sys.argv[1] |
---|
| 16 | output_filename = sys.argv[2] |
---|
| 17 | output_id = sys.argv[3] |
---|
| 18 | #where to store files that become additional output |
---|
| 19 | database_tmp_dir = sys.argv[4] |
---|
| 20 | primary_spec = sys.argv[5] |
---|
| 21 | species = sys.argv[6].split( ',' ) |
---|
| 22 | all_species = sys.argv[7].split( ',' ) |
---|
| 23 | partial = sys.argv[8] |
---|
| 24 | keep_gaps = sys.argv[9] |
---|
| 25 | out_files = {} |
---|
| 26 | |
---|
| 27 | if "None" in species: |
---|
| 28 | species = [] |
---|
| 29 | |
---|
| 30 | if primary_spec not in species: |
---|
| 31 | species.append( primary_spec ) |
---|
| 32 | if primary_spec not in all_species: |
---|
| 33 | all_species.append( primary_spec ) |
---|
| 34 | |
---|
| 35 | all_species.sort() |
---|
| 36 | for spec in species: |
---|
| 37 | if spec == primary_spec: |
---|
| 38 | out_files[ spec ] = open( output_filename, 'wb+' ) |
---|
| 39 | else: |
---|
| 40 | out_files[ spec ] = open( os.path.join( database_tmp_dir, 'primary_%s_%s_visible_interval_%s' % ( output_id, spec, spec ) ), 'wb+' ) |
---|
| 41 | out_files[ spec ].write( '#chrom\tstart\tend\tstrand\tscore\tname\t%s\n' % ( '\t'.join( all_species ) ) ) |
---|
| 42 | num_species = len( all_species ) |
---|
| 43 | |
---|
| 44 | file_in = open( input_filename, 'r' ) |
---|
| 45 | maf_reader = maf.Reader( file_in ) |
---|
| 46 | |
---|
| 47 | for i, m in enumerate( maf_reader ): |
---|
| 48 | for j, block in enumerate( maf_utilities.iter_blocks_split_by_species( m ) ): |
---|
| 49 | if len( block.components ) < num_species and partial == "partial_disallowed": continue |
---|
| 50 | sequences = {} |
---|
| 51 | for c in block.components: |
---|
| 52 | spec, chrom = maf_utilities.src_split( c.src ) |
---|
| 53 | if keep_gaps == 'remove_gaps': |
---|
| 54 | sequences[ spec ] = c.text.replace( '-', '' ) |
---|
| 55 | else: |
---|
| 56 | sequences[ spec ] = c.text |
---|
| 57 | sequences = '\t'.join( [ sequences.get( spec, '' ) for spec in all_species ] ) |
---|
| 58 | for spec in species: |
---|
| 59 | c = block.get_component_by_src_start( spec ) |
---|
| 60 | if c is not None: |
---|
| 61 | spec2, chrom = maf_utilities.src_split( c.src ) |
---|
| 62 | assert spec2 == spec, Exception( 'Species name inconsistancy found in component: %s != %s' % ( spec, spec2 ) ) |
---|
| 63 | out_files[ spec ].write( "%s\t%s\t%s\t%s\t%s\t%s\t%s\n" % ( chrom, c.forward_strand_start, c.forward_strand_end, c.strand, m.score, "%s_%s_%s" % (spec, i, j), sequences ) ) |
---|
| 64 | file_in.close() |
---|
| 65 | for file_out in out_files.values(): |
---|
| 66 | file_out.close() |
---|
| 67 | |
---|
| 68 | if __name__ == "__main__": __main__() |
---|