Mercurial > repos > blankenberg > naive_variant_caller

annotate tools/naive_variant_caller.py @ 5:8398666758e3

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Update options.
author Daniel Blankenberg <dan@bx.psu.edu>
date Tue, 04 Feb 2014 17:26:26 -0500
parents ae6edc0012ba
children d5f866df45ae
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
1 #Dan Blankenberg
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
2 import sys
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
3 import optparse
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
4
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
5 from pyBamParser.bam import Reader
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
6 from pyBamTools.genotyping.naive import VCFReadGroupGenotyper
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
7
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
8 def main():
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
9 #Parse Command Line
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
10 parser = optparse.OptionParser()
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
11 parser.add_option( '-b', '--bam', dest='bam_file', action='append', type="string", default=[], help='BAM filename, optionally index filename. Multiple allowed.' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
12 parser.add_option( '-i', '--index', dest='index_file', action='append', type="string", default=[], help='optionally index filename. Multiple allowed.' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
13 parser.add_option( '-o', '--output_vcf_filename', dest='output_vcf_filename', action='store', default = None, type="string", help='Output VCF filename' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
14 parser.add_option( '-r', '--reference_genome_filename', dest='reference_genome_filename', action='store', default = None, type="string", help='Input reference file' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
15 parser.add_option( '-v', '--variants_only', dest='variants_only', action='store_true', default = False, help='Report only sites with a possible variant allele.' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
16 parser.add_option( '-s', '--use_strand', dest='use_strand', action='store_true', default = False, help='Report counts by strand' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
17 parser.add_option( '-p', '--ploidy', dest='ploidy', action='store', type="int", default=2, help='Ploidy. Default=2.' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
18 parser.add_option( '-d', '--min_support_depth', dest='min_support_depth', action='store', type="int", default=0, help='Minimum number of reads needed to consider a REF/ALT. Default=0.' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
19 parser.add_option( '-q', '--min_base_quality', dest='min_base_quality', action='store', type="int", default=None, help='Minimum base quality.' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
20 parser.add_option( '-m', '--min_mapping_quality', dest='min_mapping_quality', action='store', type="int", default=None, help='Minimum mapping.' )
5
8398666758e3 Update options.
Daniel Blankenberg <dan@bx.psu.edu>
parents: 1
diff changeset
21 parser.add_option( '-t', '--coverage_dtype', dest='coverage_dtype', action='store', type="string", default='uint64', help='dtype to use for coverage array' )
1
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
22 parser.add_option( '--allow_out_of_bounds_positions', dest='allow_out_of_bounds_positions', action='store_true', default = False, help='Allows out of bounds positions to not throw fatal errors' )
5
8398666758e3 Update options.
Daniel Blankenberg <dan@bx.psu.edu>
parents: 1
diff changeset
23 parser.add_option( '--safe', dest='safe', action='store_true', default = False, help='Perform checks to prevent certain errors. Is slower.' )
1
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
24 parser.add_option( '--region', dest='region', action='append', type="string", default=[], help='region' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
25 (options, args) = parser.parse_args()
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
26
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
27 if len( options.bam_file ) == 0:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
28 print >>sys.stderr, 'You must provide at least one bam (-b) file.'
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
29 parser.print_help( sys.stderr )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
30 sys.exit( 1 )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
31 if options.index_file:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
32 assert len( options.index_file ) == len( options.bam_file ), "If you provide a name for an index file, you must provide the index name for all bam files."
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
33 bam_files = zip( options.bam_file, options.index_file )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
34 else:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
35 bam_files = [ ( x, ) for x in options.bam_file ]
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
36 if not options.reference_genome_filename:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
37 print >> sys.stderr, "Warning: Reference file has not been specified. Providing a reference genome is highly recommended."
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
38 if options.output_vcf_filename:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
39 out = open( options.output_vcf_filename, 'wb' )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
40 else:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
41 out = sys.stdout
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
42
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
43 regions = []
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
44 if options.region:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
45 for region in options.region:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
46 region_split = region.split( ":" )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
47 region = region_split.pop( 0 )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
48 if region_split:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
49 region_split = filter( bool, region_split[0].split( '-' ) )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
50 if region_split:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
51 if len( region_split ) != 2:
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
52 print >> sys.stderr, "You must specify both a start and an end, or only a chromosome when specifying regions."
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
53 cleanup_before_exit( tmp_dir )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
54 sys.exit( 1 )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
55 region = tuple( [ region ] + map( int, region_split ) )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
56 regions.append( region )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
57
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
58 coverage = VCFReadGroupGenotyper( map( lambda x: Reader( *x ), bam_files ), options.reference_genome_filename, dtype=options.coverage_dtype,
5
8398666758e3 Update options.
Daniel Blankenberg <dan@bx.psu.edu>
parents: 1
diff changeset
59 min_support_depth=options.min_support_depth, min_base_quality=options.min_base_quality,
8398666758e3 Update options.
Daniel Blankenberg <dan@bx.psu.edu>
parents: 1
diff changeset
60 min_mapping_quality=options.min_mapping_quality, restrict_regions=regions, use_strand=options.use_strand,
8398666758e3 Update options.
Daniel Blankenberg <dan@bx.psu.edu>
parents: 1
diff changeset
61 allow_out_of_bounds_positions=options.allow_out_of_bounds_positions, safe=options.safe )
1
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
62 for line in coverage.iter_vcf( ploidy=options.ploidy, variants_only=options.variants_only ):
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
63 out.write( "%s\n" % line )
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
64 out.close()
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
65
ae6edc0012ba Populate naive_variant_caller repository.
Daniel Blankenberg <dan@bx.psu.edu>
parents:
diff changeset
66 if __name__ == "__main__": main()