def filter_snvs(in_fhand, out_fhand, filters, filtered_fhand=None,
log_fhand=None, reader_kwargs=None):
'''IT filters an input vcf.
The input fhand has to be uncompressed. The original file could be a
gzipped file, but in that case it has to be opened with gzip.open before
sending it to this function.
'''
if reader_kwargs is None:
reader_kwargs = {}
# The input fhand to this function cannot be compressed
reader_kwargs.update({'compressed': False,
'filename': 'pyvcf_bug_workaround'})
reader = VCFReader(in_fhand, **reader_kwargs)
template_reader = VCFReader(StringIO(reader.header))
writer = VCFWriter(out_fhand, template_reader=template_reader)
if filtered_fhand:
filtered_writer = VCFWriter(filtered_fhand,
template_reader=template_reader)
else:
filtered_writer = None
packets = group_in_filter_packets(reader.parse_snvs(),
SNPS_PER_FILTER_PACKET)
tot_snps = 00.01
passed_snps = OrderedDict()
broken_pipe = False
for packet in packets:
tot_snps += len(packet[PASSED]) + len(packet[FILTERED_OUT])
for filter_ in filters:
packet = filter_(packet)
filter_name = filter_.__class__.__name__
if filter_name not in passed_snps:
passed_snps[filter_name] = 0
passed_snps[filter_name] += len(packet[PASSED])
for snv in packet[PASSED]:
if not _safe_write_snv(writer, snv):
broken_pipe = True
break
if filtered_writer:
for snv in packet[FILTERED_OUT]:
if not _safe_write_snv(filtered_writer, snv):
broken_pipe = True
break
if broken_pipe:
break
if log_fhand:
_write_log(log_fhand, tot_snps, passed_snps)
writer.flush()
def test_het_unknown(self):
vcf = '''#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT 1 2 3 4 5 6 7 8
20\t2\t.\tG\tA\t29\tPASS\tNS=3\tGT\t0/0\t0/0\t0/0\t0/0\t1/1\t1/1\t./.\t1/.\t
'''
vcf = StringIO(VCF_HEADER + vcf)
reader = VCFReader(vcf)
snps = list(reader.parse_snvs())
snp = snps[0]
expected = [[0, 0], [0, 0], [0, 0], [0, 0], [1, 1], [1, 1], [],
[1, None]]
assert [call.int_alleles for call in snps[0].calls] == expected
assert snp.num_called == 7
out_fhand = StringIO()
writer = VCFWriter(out_fhand, reader)
for snv in snps:
writer.write_snv(snv)
assert '1/1\t./.\t1/.' in out_fhand.getvalue()
def test_vcf_writer(self):
varscan = open(join(TEST_DATA_DIR, 'vari_filter.vcf'))
reader = VCFReader(fhand=varscan)
out_fhand = NamedTemporaryFile()
writer = VCFWriter(out_fhand, reader)
for snv in reader.parse_snvs():
writer.write_snv(snv)
writer.flush()
assert 'CUUC00027_TC01' in open(out_fhand.name).read()
writer.close()
def run_genotype_filters(in_fhand, out_fhand, gt_filters, reader_kwargs=None):
if reader_kwargs is None:
reader_kwargs = {}
reader_kwargs['filename'] = 'pyvcf_bug_workaround'
reader_kwargs['compressed'] = False
reader = VCFReader(in_fhand, **reader_kwargs)
templa_reader = VCFReader(StringIO(reader.header))
writer = VCFWriter(out_fhand, template_reader=templa_reader)
for snv in reader.parse_snvs():
for mapper in gt_filters:
snv = mapper(snv)
try:
writer.write_snv(snv)
except IOError, error:
# The pipe could be already closed
if 'Broken pipe' in str(error):
break
else:
raise