def deepvariant_header(contigs, sample_names):
"""Returns a VcfHeader used for writing VCF output.
This function fills out the FILTER, INFO, FORMAT, and extra header information
created by the DeepVariant pipeline using consistent fields that DeepVariant
creates. The `contigs` and `sample_names` fields are unique depending on the
input data used, so are required inputs.
Args:
contigs: list(ContigInfo). The list of contigs on which variants were
called.
sample_names: list(str). The list of samples present in the run.
Returns:
A nucleus.genomics.v1.VcfHeader proto with known fixed headers and the given
samples and contigs populated.
"""
version = variants_pb2.VcfExtra(key='DeepVariant_version',
value=DEEP_VARIANT_VERSION)
return variants_pb2.VcfHeader(
fileformat='VCFv4.2',
filters=[
vcf_constants.reserved_filter_field(DEEP_VARIANT_PASS),
variants_pb2.VcfFilterInfo(
id=DEEP_VARIANT_REF_FILTER,
description='Genotyping model thinks this site is reference.'),
variants_pb2.VcfFilterInfo(
id=DEEP_VARIANT_QUAL_FILTER,
description='Confidence in this variant being real is below '
'calling threshold.'),
],
infos=[
vcf_constants.reserved_info_field('END'),
],
formats=[
vcf_constants.reserved_format_field('GT'),
vcf_constants.reserved_format_field('GQ'),
vcf_constants.reserved_format_field('DP'),
variants_pb2.VcfFormatInfo(
id=DEEP_VARIANT_MIN_DP_FORMAT,
number='1',
type='Integer',
description='Minimum DP observed within the GVCF block.'),
vcf_constants.reserved_format_field('AD'),
variants_pb2.VcfFormatInfo(
id=DEEP_VARIANT_VAF_FORMAT,
number='A',
type='Float',
description='Variant allele fractions.'),
vcf_constants.reserved_format_field('PL'),
],
contigs=contigs,
sample_names=sample_names,
extras=[version])
def test_vcf_header(self):
header = self.sites_reader.header
expected1 = variants_pb2.VcfStructuredExtra(
key='ALT',
fields=[
variants_pb2.VcfExtra(key='ID', value='NON_REF'),
variants_pb2.VcfExtra(
key='Description',
value='Represents any possible alternative allele at th'
'is location')
])
expected2 = variants_pb2.VcfStructuredExtra(
key='META',
fields=[
variants_pb2.VcfExtra(key='ID', value='TESTMETA'),
variants_pb2.VcfExtra(key='Description', value='blah')
])
self.assertLen(header.structured_extras, 2)
self.assertEqual(header.structured_extras[1], expected2)
self.assertEqual(header.structured_extras[0], expected1)
def deepvariant_header(contigs,
sample_names,
add_info_candidates=False,
include_med_dp=True):
"""Returns a VcfHeader used for writing VCF output.
This function fills out the FILTER, INFO, FORMAT, and extra header information
created by the DeepVariant pipeline using consistent fields that DeepVariant
creates. The `contigs` and `sample_names` fields are unique depending on the
input data used, so are required inputs.
Args:
contigs: list(ContigInfo). The list of contigs on which variants were
called.
sample_names: list(str). The list of samples present in the run.
add_info_candidates: Adds the 'CANDIDATES' info field for
debugging purposes.
include_med_dp: boolean. If True, we will include MED_DP.
Returns:
A nucleus.genomics.v1.VcfHeader proto with known fixed headers and the given
samples and contigs populated.
"""
version = variants_pb2.VcfExtra(
key='DeepVariant_version', value=DEEP_VARIANT_VERSION)
info_fields = [
vcf_constants.reserved_info_field('END'),
]
formats = [
vcf_constants.reserved_format_field('GT'),
vcf_constants.reserved_format_field('GQ'),
vcf_constants.reserved_format_field('DP'),
variants_pb2.VcfFormatInfo(
id=DEEP_VARIANT_MIN_DP_FORMAT,
number='1',
type='Integer',
description='Minimum DP observed within the GVCF block.'),
vcf_constants.reserved_format_field('AD'),
variants_pb2.VcfFormatInfo(
id=DEEP_VARIANT_VAF_FORMAT,
number='A',
type='Float',
description='Variant allele fractions.'),
vcf_constants.reserved_format_field('PL'),
]
if add_info_candidates:
info_fields.append(
variants_pb2.VcfInfo(
id='CANDIDATES',
number='1',
type=vcf_constants.STRING_TYPE,
description='pipe-delimited candidate alleles.'))
if include_med_dp:
formats.append(
variants_pb2.VcfFormatInfo(
id=DEEP_VARIANT_MED_DP_FORMAT,
number='1',
type='Integer',
description='Median DP observed within the GVCF block '
'rounded to the nearest integer.'))
return variants_pb2.VcfHeader(
fileformat='VCFv4.2',
filters=[
vcf_constants.reserved_filter_field(DEEP_VARIANT_PASS),
variants_pb2.VcfFilterInfo(
id=DEEP_VARIANT_REF_FILTER,
description='Genotyping model thinks this site is reference.'),
variants_pb2.VcfFilterInfo(
id=DEEP_VARIANT_QUAL_FILTER,
description='Confidence in this variant being real is below '
'calling threshold.'),
variants_pb2.VcfFilterInfo(
id=DEEP_VARIANT_NO_CALL,
description='Site has depth=0 resulting in no call.'),
],
infos=info_fields,
formats=formats,
contigs=contigs,
sample_names=sample_names,
extras=[version])