def main():
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('vcf', help='Module 04 genotyped PESR vcf')
parser.add_argument('allosome_contigs_file')
parser.add_argument('famfile', type=argparse.FileType('r'))
parser.add_argument('case_sample')
parser.add_argument('min_size',
help='minumum size at which to apply conversions',
type=int)
parser.add_argument('-o',
'--outfile',
help='Output file [default: stdout]')
args = parser.parse_args()
if args.vcf in '- stdin'.split():
vcf = pysam.VariantFile(sys.stdin)
else:
vcf = pysam.VariantFile(args.vcf)
header = vcf.header
case_sample = args.case_sample
min_size = args.min_size
if args.outfile is None:
fout = pysam.VariantFile(sys.stdout, 'w', header=header)
else:
out = args.outfile
fout = pysam.VariantFile(out, 'w', header=header)
allosome_contigs = read_contigs_list(args.allosome_contigs_file)
fam = parse_famfile(args.famfile)
case_sample_sex = fam.samples[args.case_sample].sex
samples_by_sex = {
'1': [s for s in fam.samples if fam.samples[s].sex == '1'],
'2': [s for s in fam.samples if fam.samples[s].sex == '2']
}
for record in vcf:
contig = record.contig
svtype = record.info['SVTYPE']
if (svtype == 'DEL'
or svtype == 'DUP') and record.info['SVLEN'] >= min_size:
if contig not in allosome_contigs:
if not has_depth_support_autosome(
record, case_sample) and has_sr_or_pe_support(
record, case_sample):
record.info['SVTYPE'] = 'BND'
record.alts = ['<BND>']
else:
if not has_depth_support_allosome(record, case_sample, samples_by_sex[case_sample_sex]) \
and has_sr_or_pe_support(record, case_sample):
record.info['SVTYPE'] = 'BND'
record.alts = ['<BND>']
fout.write(record)
def main():
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('vcf')
parser.add_argument('famfile', type=argparse.FileType('r'))
parser.add_argument('var_fout', type=argparse.FileType('w'))
parser.add_argument('obs_fout', type=argparse.FileType('w'))
args = parser.parse_args()
vcf = pysam.VariantFile(args.vcf)
fam = parse_famfile(args.famfile)
scraper = StatsScraper(vcf, fam, args.var_fout, args.obs_fout)
scraper.scrape()
def main():
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('vcf')
parser.add_argument('filtered',
type=argparse.FileType('r'),
help='De novo filter results')
parser.add_argument('famfile', type=argparse.FileType('r'))
parser.add_argument('fout', help='Filtered VCF')
args = parser.parse_args()
vcf = pysam.VariantFile(args.vcf)
fam = parse_famfile(args.famfile)
fout = sys.stdout if args.fout in 'stdout -'.split() else args.fout
fout = pysam.VariantFile(fout, 'w', header=vcf.header)
filter_dn_variants(vcf, args.filtered, fam, fout)
def main():
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('metrics', type=argparse.FileType('r'))
parser.add_argument('cutoffs')
parser.add_argument('famfile', type=argparse.FileType('r'))
parser.add_argument('fout', type=argparse.FileType('w'))
args = parser.parse_args()
fam = parse_famfile(args.famfile)
metrics = metric_parser(args.metrics, fam)
cutoffs = pd.read_table(args.cutoffs)
cutoffs = format_cutoffs(cutoffs)
fmt = '{0}\t{1}\t{2}\t{3}\n'
for name, family, samples, support in filter_denovo(metrics, fam, cutoffs):
args.fout.write(fmt.format(name, family, samples, support))
def main():
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('vcf')
parser.add_argument('famfile', type=argparse.FileType('r'))
parser.add_argument('-c', '--countfile', required=True)
parser.add_argument('-d', '--discfile', required=True)
parser.add_argument('--discfile-index')
parser.add_argument('--countfile-index')
parser.add_argument('--background', type=int, default=160)
parser.add_argument('--max-parents', type=float, default=10)
parser.add_argument('petest', type=argparse.FileType('w'), help='fout')
parser.add_argument('srtest', type=argparse.FileType('w'), help='fout')
args = parser.parse_args()
vcf = pysam.VariantFile(args.vcf)
fam = parse_famfile(args.famfile)
if args.discfile_index is None:
discfile = pysam.TabixFile(args.discfile, parser=pysam.asTuple())
else:
discfile = pysam.TabixFile(args.discfile,
index=args.discfile_index,
parser=pysam.asTuple())
if args.countfile_index is None:
countfile = pysam.TabixFile(args.countfile, parser=pysam.asTuple())
else:
countfile = pysam.TabixFile(args.countfile,
index=args.countfile_index,
parser=pysam.asTuple())
header = 'name sample log_pval called_median bg_median'.split()
args.petest.write('\t'.join(header) + '\n')
header = 'name sample coord pos log_pval called_median bg_median'.split()
args.srtest.write('\t'.join(header) + '\n')
runner = DenovoTestRunner(vcf, fam, countfile, discfile, args.petest,
args.srtest, args.background, args.max_parents)
runner.run()
def main():
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('vcf')
parser.add_argument('famfile', type=argparse.FileType('r'))
parser.add_argument('fout')
parser.add_argument('--max-parents', type=int, default=10)
args = parser.parse_args()
if args.vcf in '- stdin'.split():
vcf = pysam.VariantFile(sys.stdin)
else:
vcf = pysam.VariantFile(args.vcf)
if args.fout in '- stdout'.split():
fout = pysam.VariantFile(sys.stdout, 'w', header=vcf.header)
else:
fout = pysam.VariantFile(args.fout, 'w', header=vcf.header)
fam = parse_famfile(args.famfile)
for record in filter_denovo_records(vcf, fam, args.max_parents):
fout.write(record)