def normalize(ref_genome, reader, writer, maxIndelLen=50, cleanOnly=False):
for record in reader:
if (record.FILTER != [] and record.FILTER != "."
and record.FILTER != "PASS" and record.FILTER != None
or genotype(record) == "0/0"):
continue # record filtered
if (genotype(record) == "." and not is_sv(record, maxIndelLen)):
continue # filter variants without genotype information unless they are structural variants
pos = record.POS - 1
contig = record.CHROM
ref = str(record.REF.upper())
alts = map(lambda a: str(a).upper(), record.ALT)
all_alleles = alts + [ref]
redundancy = find_redundancy(all_alleles)
if cleanOnly:
write(record, pos, ref, alts, writer)
continue
if redundancy:
def chop(allele):
return allele[:-redundancy]
ref = chop(ref)
alts = map(chop, alts)
pos, ref, alts = left_normalize(ref_genome, record.CHROM, pos, ref,
alts)
write(record, pos, ref, alts, writer)
prev_end = pos + len(ref)
prev_contig = contig
def keep_variant(record,maxIndelLen=50):
if ( record.FILTER != [] and record.FILTER != "." and record.FILTER != "PASS" and record.FILTER != None or genotype(record) == "0/0"):
return False # record filtered
if ( genotype(record) == "." and not is_sv(record, maxIndelLen)):
return False # filter variants without genotype information unless they are structural variants
if ( "N" in record.REF or any(map(lambda a: "N" in str(a), record.ALT))):
return False # filter out ref/alts with N base
return True
def keep_variant(record, maxIndelLen=50):
if (record.FILTER != [] and record.FILTER != "."
and record.FILTER != "PASS" and record.FILTER != None
or genotype(record) == "0/0"):
return False # record filtered
if (genotype(record) == "." and not is_sv(record, maxIndelLen)):
return False # filter variants without genotype information unless they are structural variants
if ("N" in record.REF or any(map(lambda a: "N" in str(a), record.ALT))):
return False # filter out ref/alts with N base
return True
def keep_variant(record,maxIndelLen=50):
if ( record.FILTER != [] and record.FILTER != "." and record.FILTER != "PASS" and record.FILTER != None or genotype(record) == "0/0"):
return False # record filtered
if ( genotype(record) == "." and not is_sv(record, maxIndelLen)):
return False # filter variants without genotype information unless they are structural variants
return True
def keep_variant(record,maxIndelLen=50):
if ( record.FILTER != [] and record.FILTER != "." and record.FILTER != "PASS" and record.FILTER != None or genotype(record) == "0/0"):
return False # record filtered
if ( genotype(record) == "." and not is_sv(record, maxIndelLen)):
return False # filter variants without genotype information unless they are structural variants
return True
