locus = int(r[snvheaders[1]].strip())
ref = r[snvheaders[2]].strip()
alt = r[snvheaders[3]].strip()
if r.get('INFO:INDEL'):
continue
if len(ref) != 1:
continue
if not re.search(r'^[ACGT](,[ACGT])*$', alt):
continue
# for h in r:
# if r.get(h):
# usedsnvheaders.add(h)
snvkey = (filename, chr, locus, ref, alt)
if snvkey not in snvdata:
snvdata[snvkey] = r
progress.update()
progress.done()
chrreg = ChromLabelRegistry()
for snvfile in snvchroms:
chrreg.add_labels(snvfile, snvchroms[snvfile])
snvdata1 = {}
for (sf, chr, locus, ref, alt), r in snvdata.iteritems():
chrom = chrreg.label2chrom(sf, chr)
assert (chrom)
snvkey = (chrom, locus, ref, alt)
if snvkey not in snvdata1:
snvdata1[snvkey] = (chrom, locus, ref, alt, r)
ref = r[snvheaders[2]].strip()
alt = r[snvheaders[3]].strip()
if r.get('INFO:INDEL'):
continue
if len(ref) != 1:
continue
if not re.search(r'^[ACGT](,[ACGT])*$', alt):
continue
# for h in r:
# if r.get(h):
# usedsnvheaders.add(h)
snvkey = (filename, chr, locus, ref, alt)
if snvkey not in snvdata:
snvdata[snvkey] = r
progress.update()
progress.done()
chrreg = ChromLabelRegistry()
for snvfile in snvchroms:
chrreg.add_labels(snvfile,snvchroms[snvfile])
snvdata1 = {}
for (sf, chr, locus, ref, alt), r in snvdata.iteritems():
chrom = chrreg.label2chrom(sf,chr)
assert(chrom)
snvkey = (chrom,locus,ref,alt)
if snvkey not in snvdata1:
snvdata1[snvkey] = (chrom,locus,ref,alt,r)
