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])
reference_pb2.ContigInfo(name='chr15', pos_in_fasta=14, n_bases=101991189),
reference_pb2.ContigInfo(name='chr16', pos_in_fasta=15, n_bases=90338345),
reference_pb2.ContigInfo(name='chr17', pos_in_fasta=16, n_bases=83257441),
reference_pb2.ContigInfo(name='chr18', pos_in_fasta=17, n_bases=80373285),
reference_pb2.ContigInfo(name='chr19', pos_in_fasta=18, n_bases=58617616),
reference_pb2.ContigInfo(name='chr20', pos_in_fasta=19, n_bases=64444167),
reference_pb2.ContigInfo(name='chr21', pos_in_fasta=20, n_bases=46709983),
reference_pb2.ContigInfo(name='chr22', pos_in_fasta=21, n_bases=50818468),
reference_pb2.ContigInfo(name='chrX', pos_in_fasta=22, n_bases=156040895),
reference_pb2.ContigInfo(name='chrY', pos_in_fasta=23, n_bases=57227415),
reference_pb2.ContigInfo(name='chrM', pos_in_fasta=24, n_bases=16569),
]
# pylint: disable=line-too-long
expected_samples_filters = [
variants_pb2.VcfFilterInfo(id='PASS', description='All filters passed'),
variants_pb2.VcfFilterInfo(id='LowQual', description='Low quality'),
variants_pb2.VcfFilterInfo(
id='VQSRTrancheINDEL95.00to96.00',
description=
'Truth sensitivity tranche level for INDEL model at VQS Lod: 0.9364 <= x < 1.0415'
),
variants_pb2.VcfFilterInfo(
id='VQSRTrancheINDEL96.00to97.00',
description=
'Truth sensitivity tranche level for INDEL model at VQS Lod: 0.8135 <= x < 0.9364'
),
variants_pb2.VcfFilterInfo(
id='VQSRTrancheINDEL97.00to99.00',
description=
'Truth sensitivity tranche level for INDEL model at VQS Lod: 0.323 <= x < 0.8135'
def test_writing_canned_variants(self):
"""Tests writing all the variants that are 'canned' in our tfrecord file."""
# This file is in the TF record format
tfrecord_file = test_utils.genomics_core_testdata(
'test_samples.vcf.golden.tfrecord')
writer_options = variants_pb2.VcfWriterOptions()
header = variants_pb2.VcfHeader(
contigs=[
reference_pb2.ContigInfo(name='chr1', n_bases=248956422),
reference_pb2.ContigInfo(name='chr2', n_bases=242193529),
reference_pb2.ContigInfo(name='chr3', n_bases=198295559),
reference_pb2.ContigInfo(name='chrX', n_bases=156040895)
],
sample_names=['NA12878_18_99'],
filters=[
variants_pb2.VcfFilterInfo(id='PASS',
description='All filters passed'),
variants_pb2.VcfFilterInfo(id='LowQual', description=''),
variants_pb2.VcfFilterInfo(id='VQSRTrancheINDEL95.00to96.00'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheINDEL96.00to97.00'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheINDEL97.00to99.00'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheINDEL99.00to99.50'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheINDEL99.50to99.90'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheINDEL99.90to99.95'),
variants_pb2.VcfFilterInfo(
id='VQSRTrancheINDEL99.95to100.00+'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheINDEL99.95to100.00'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheSNP99.50to99.60'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheSNP99.60to99.80'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheSNP99.80to99.90'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheSNP99.90to99.95'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheSNP99.95to100.00+'),
variants_pb2.VcfFilterInfo(id='VQSRTrancheSNP99.95to100.00'),
],
infos=[
variants_pb2.VcfInfo(
id='END',
number='1',
type='Integer',
description='Stop position of the interval')
],
formats=[
variants_pb2.VcfFormatInfo(id='GT',
number='1',
type='String',
description='Genotype'),
variants_pb2.VcfFormatInfo(id='GQ',
number='1',
type='Integer',
description='Genotype Quality'),
variants_pb2.VcfFormatInfo(
id='DP',
number='1',
type='Integer',
description='Read depth of all passing filters reads.'),
variants_pb2.VcfFormatInfo(
id='MIN_DP',
number='1',
type='Integer',
description='Minimum DP observed within the GVCF block.'),
variants_pb2.VcfFormatInfo(
id='AD',
number='R',
type='Integer',
description=
'Read depth of all passing filters reads for each allele.'
),
variants_pb2.VcfFormatInfo(
id='VAF',
number='A',
type='Float',
description='Variant allele fractions.'),
variants_pb2.VcfFormatInfo(
id='PL',
number='G',
type='Integer',
description='Genotype likelihoods, Phred encoded'),
],
)
variant_records = list(
io_utils.read_tfrecords(tfrecord_file, proto=variants_pb2.Variant))
out_fname = test_utils.test_tmpfile('output.vcf')
with vcf_writer.VcfWriter.to_file(out_fname, header,
writer_options) as writer:
for record in variant_records[:5]:
writer.write(record)
# Check: are the variants written as expected?
# pylint: disable=line-too-long
expected_vcf_content = [
'##fileformat=VCFv4.2\n',
'##FILTER=<ID=PASS,Description="All filters passed">\n',
'##FILTER=<ID=LowQual,Description="">\n',
'##FILTER=<ID=VQSRTrancheINDEL95.00to96.00,Description="">\n',
'##FILTER=<ID=VQSRTrancheINDEL96.00to97.00,Description="">\n',
'##FILTER=<ID=VQSRTrancheINDEL97.00to99.00,Description="">\n',
'##FILTER=<ID=VQSRTrancheINDEL99.00to99.50,Description="">\n',
'##FILTER=<ID=VQSRTrancheINDEL99.50to99.90,Description="">\n',
'##FILTER=<ID=VQSRTrancheINDEL99.90to99.95,Description="">\n',
'##FILTER=<ID=VQSRTrancheINDEL99.95to100.00+,Description="">\n',
'##FILTER=<ID=VQSRTrancheINDEL99.95to100.00,Description="">\n',
'##FILTER=<ID=VQSRTrancheSNP99.50to99.60,Description="">\n',
'##FILTER=<ID=VQSRTrancheSNP99.60to99.80,Description="">\n',
'##FILTER=<ID=VQSRTrancheSNP99.80to99.90,Description="">\n',
'##FILTER=<ID=VQSRTrancheSNP99.90to99.95,Description="">\n',
'##FILTER=<ID=VQSRTrancheSNP99.95to100.00+,Description="">\n',
'##FILTER=<ID=VQSRTrancheSNP99.95to100.00,Description="">\n',
'##INFO=<ID=END,Number=1,Type=Integer,Description="Stop position of '
'the interval">\n',
'##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">\n',
'##FORMAT=<ID=GQ,Number=1,Type=Integer,Description="Genotype Quality">\n',
'##FORMAT=<ID=DP,Number=1,Type=Integer,Description="Read depth of all '
'passing filters reads.">\n',
'##FORMAT=<ID=MIN_DP,Number=1,Type=Integer,Description="Minimum DP '
'observed within the GVCF block.">\n',
'##FORMAT=<ID=AD,Number=R,Type=Integer,Description="Read depth of all '
'passing filters reads for each allele.">\n',
'##FORMAT=<ID=VAF,Number=A,Type=Float,Description=\"Variant allele '
'fractions.">\n',
'##FORMAT=<ID=PL,Number=G,Type=Integer,Description="Genotype '
'likelihoods, Phred encoded">\n',
'##contig=<ID=chr1,length=248956422>\n',
'##contig=<ID=chr2,length=242193529>\n',
'##contig=<ID=chr3,length=198295559>\n',
'##contig=<ID=chrX,length=156040895>\n',
'#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tNA12878_18_99\n',
'chr1\t13613\t.\tT\tA\t39.88\tVQSRTrancheSNP99.90to99.95\t.\tGT:GQ:DP:AD:PL\t0/1:16:4:1,3:68,0,16\n',
'chr1\t13813\t.\tT\tG\t90.28\tPASS\t.\tGT:GQ:DP:AD:PL\t1/1:9:3:0,3:118,9,0\n',
'chr1\t13838\trs28428499\tC\tT\t62.74\tPASS\t.\tGT:GQ:DP:AD:PL\t1/1:6:2:0,2:90,6,0\n',
'chr1\t14397\trs756427959\tCTGT\tC\t37.73\tPASS\t.\tGT:GQ:DP:AD:PL\t0/1:75:5:3,2:75,0,152\n',
'chr1\t14522\t.\tG\tA\t49.77\tVQSRTrancheSNP99.60to99.80\t.\tGT:GQ:DP:AD:PL\t0/1:78:10:6,4:78,0,118\n'
]
# pylint: enable=line-too-long
with gfile.GFile(out_fname, 'r') as f:
self.assertEqual(f.readlines(), expected_vcf_content)
# Older symbolic alt allele, similar in meaning to gVCF alt allele
SYMBOLIC_ALT_ALLELE = '<NON_REF>'
# The replacement field used for missing data.
MISSING_FIELD = '.'
# Valid types for INFO and FORMAT fields, as per the VCF 4.3 spec.
CHARACTER_TYPE = 'Character'
FLAG_TYPE = 'Flag'
FLOAT_TYPE = 'Float'
INTEGER_TYPE = 'Integer'
STRING_TYPE = 'String'
# Reserved FILTER field definitions.
RESERVED_FILTER_FIELDS = [
variants_pb2.VcfFilterInfo(id='PASS', description='All filters passed'),
]
# Reserved INFO field definitions, as per the VCF 4.3 spec.
RESERVED_INFO_FIELDS = [
variants_pb2.VcfInfo(id='AA',
number='1',
type=STRING_TYPE,
description='Ancestral allele'),
variants_pb2.VcfInfo(id='AC',
number='A',
type=INTEGER_TYPE,
description='Allele count in genotypes, for each ALT '
'allele, in the same order as listed'),
variants_pb2.VcfInfo(id='AD',
number='R',
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])
