def setUp(self):
self.variant = Variant(chrom='chr1',
pos='14752',
id_='.',
ref='G',
alt='A',
qual='.',
filter_='weak_evidence',
info='DP=236;FOO',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:113,4:0.063',
'TUMOR': '0/1:113,4:0.063'
})
def parse_vcf(stream: Iterable[str]) -> VCF:
header: List[str] = []
sample_names: List[str] = []
variants: List[Variant] = []
for row in stream:
if row.startswith('##'):
header.append(row.rstrip('\n'))
elif row.startswith('#'):
sample_names = row.rstrip('\n').split('\t')[9:]
else:
chrom, pos, id_, ref, alt, qual, filter_, info, format_, *samples = row.rstrip(
'\n').split('\t')
variants.append(
Variant(chrom=chrom,
pos=pos,
id_=id_,
ref=ref,
alt=alt,
qual=qual,
filter_=filter_,
info=info,
format_=format_,
samples=dict(zip(sample_names, samples))))
return VCF(header=header, sample_names=sample_names, variants=variants)
def test_variant_filter_status(self):
non_filtered_variant = Variant(chrom='chr1',
pos='11117039',
id_='.',
ref='C',
alt='A',
qual='.',
filter_='PASS',
info='DP=264',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:128,0:9.577e-03',
'TUMOR': '0/1:99,33:0.204'
})
self.assertTrue(self.variant.is_filtered())
self.assertFalse(non_filtered_variant.is_filtered())
def test_triager_multiple_variants_multiple_evidence(self):
vcf = VCF(
header=[],
sample_names=[],
variants=[
Variant(chrom='chr1', pos='100', id_='.', ref='A', alt='T', qual='.', filter_='PASS', info='', format_='', samples={}),
Variant(chrom='chr2', pos='500', id_='.', ref='G', alt='GATATA', qual='.', filter_='weak_evidence', info='', format_='', samples={}),
Variant(chrom='chr3', pos='7500', id_='.', ref='ACT', alt='A', qual='.', filter_='germline', info='', format_='', samples={}),
Variant(chrom='chr4', pos='1980', id_='.', ref='GGGG', alt='G', qual='.', filter_='haplotype', info='', format_='', samples={}),
Variant(chrom='chr5', pos='5700', id_='.', ref='CTCTCT', alt='C', qual='.', filter_='slippage', info='', format_='', samples={}),
]
)
triager = Triager(evidence={
'X': [
Variant(chrom='chr2', pos='500', id_='.', ref='G', alt='GATATA', qual='.', filter_='PASS', info='', format_='', samples={}),
Variant(chrom='chr3', pos='7500', id_='.', ref='ACT', alt='A', qual='.', filter_='contamination', info='', format_='', samples={})
],
'Y': [
Variant(chrom='chr2', pos='500', id_='.', ref='G', alt='GATATA', qual='.', filter_='PASS', info='', format_='', samples={}),
Variant(chrom='chr4', pos='1980', id_='.', ref='GGGG', alt='G', qual='.', filter_='PASS', info='', format_='', samples={}),
]
})
triager.triage(vcf=vcf)
self.assertFalse(vcf.variants[0].is_filtered())
self.assertIs(vcf.variants[0].get_info('VTSO'), False)
self.assertIs(vcf.variants[0].get_info('VTOF'), False)
self.assertFalse(vcf.variants[1].is_filtered())
self.assertEqual(vcf.variants[1].get_info('VTSO'), 'X,Y')
self.assertEqual(vcf.variants[1].get_info('VTOF'), 'weak_evidence')
self.assertTrue(vcf.variants[2].is_filtered())
self.assertIs(vcf.variants[2].get_info('VTSO'), False)
self.assertIs(vcf.variants[2].get_info('VTOF'), False)
self.assertFalse(vcf.variants[3].is_filtered())
self.assertEqual(vcf.variants[3].get_info('VTSO'), 'Y')
self.assertEqual(vcf.variants[3].get_info('VTOF'), 'haplotype')
self.assertTrue(vcf.variants[4].is_filtered())
self.assertIs(vcf.variants[4].get_info('VTSO'), False)
self.assertIs(vcf.variants[4].get_info('VTOF'), False)
def test_triage_filtered_variant_which_passes_in_evidence(self):
vcf = VCF(
header=[],
sample_names=[],
variants=[
Variant(chrom='chr1', pos='100', id_='.', ref='A', alt='T', qual='.', filter_='clustered_events', info='', format_='', samples={})
]
)
triager = Triager(evidence={
'X': [
Variant(chrom='chr1', pos='100', id_='.', ref='A', alt='T', qual='.', filter_='PASS', info='', format_='', samples={})
]
})
triager.triage(vcf=vcf)
self.assertFalse(vcf.variants[0].is_filtered())
self.assertEqual(vcf.variants[0].get_info('VTSO'), 'X')
self.assertEqual(vcf.variants[0].get_info('VTOF'), 'clustered_events')
def test_creating_multiallelic_variant_raises_exception(self):
with self.assertRaises(VariantException):
Variant(chrom='chr1',
pos='8013449',
id_='.',
ref='C',
alt='G,A',
qual='.',
filter_='clustered_events;multiallelic',
info='DP=238',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:113,2,0:0.030',
'TUMOR': '0/1/2:59,2,56:0.029,0.440'
})
def test_variant_inequality(self):
other_variant = Variant(chrom='chr1',
pos='1625272',
id_='.',
ref='GC',
alt='G',
qual='.',
filter_='clustered_events;weak_evidence',
info='DP=455;STR',
format_='GT:AD:AF',
samples={
'NORMAL': '0|0:224,3:0.015',
'TUMOR': '0|1:225,3:0.015'
})
self.assertNotEqual(self.variant, other_variant)
def test_variant_equality(self):
same_variant = Variant(chrom='chr1',
pos='14752',
id_='.',
ref='G',
alt='A',
qual='.',
filter_='weak_evidence',
info='DP=236',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:113,4:0.063',
'TUMOR': '0/1:113,4:0.063'
})
self.assertEqual(self.variant, same_variant)
def test_variant_equality_different_filter_status(self):
same_variant_but_not_filtered = Variant(chrom='chr1',
pos='14752',
id_='.',
ref='G',
alt='A',
qual='.',
filter_='PASS',
info='DP=236',
format_='GT:AD:AF',
samples={
'NORMAL':
'0/0:113,4:0.063',
'TUMOR': '0/1:113,4:0.063'
})
self.assertEqual(self.variant, same_variant_but_not_filtered)
class TestVariant(unittest.TestCase):
def setUp(self):
self.variant = Variant(chrom='chr1',
pos='14752',
id_='.',
ref='G',
alt='A',
qual='.',
filter_='weak_evidence',
info='DP=236;FOO',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:113,4:0.063',
'TUMOR': '0/1:113,4:0.063'
})
def test_variant_equality(self):
same_variant = Variant(chrom='chr1',
pos='14752',
id_='.',
ref='G',
alt='A',
qual='.',
filter_='weak_evidence',
info='DP=236',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:113,4:0.063',
'TUMOR': '0/1:113,4:0.063'
})
self.assertEqual(self.variant, same_variant)
def test_variant_equality_different_filter_status(self):
same_variant_but_not_filtered = Variant(chrom='chr1',
pos='14752',
id_='.',
ref='G',
alt='A',
qual='.',
filter_='PASS',
info='DP=236',
format_='GT:AD:AF',
samples={
'NORMAL':
'0/0:113,4:0.063',
'TUMOR': '0/1:113,4:0.063'
})
self.assertEqual(self.variant, same_variant_but_not_filtered)
def test_variant_inequality(self):
other_variant = Variant(chrom='chr1',
pos='1625272',
id_='.',
ref='GC',
alt='G',
qual='.',
filter_='clustered_events;weak_evidence',
info='DP=455;STR',
format_='GT:AD:AF',
samples={
'NORMAL': '0|0:224,3:0.015',
'TUMOR': '0|1:225,3:0.015'
})
self.assertNotEqual(self.variant, other_variant)
def test_variant_comparison_with_non_variant_object(self):
not_a_variant = 'chr1 14752 . G A . weak_evidence DP=236 GT:AD:AF 0/0:113,4:0.063 0/1:113,4:0.063'
self.assertNotEqual(self.variant, not_a_variant)
def test_variant_filter_status(self):
non_filtered_variant = Variant(chrom='chr1',
pos='11117039',
id_='.',
ref='C',
alt='A',
qual='.',
filter_='PASS',
info='DP=264',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:128,0:9.577e-03',
'TUMOR': '0/1:99,33:0.204'
})
self.assertTrue(self.variant.is_filtered())
self.assertFalse(non_filtered_variant.is_filtered())
def test_creating_multiallelic_variant_raises_exception(self):
with self.assertRaises(VariantException):
Variant(chrom='chr1',
pos='8013449',
id_='.',
ref='C',
alt='G,A',
qual='.',
filter_='clustered_events;multiallelic',
info='DP=238',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:113,2,0:0.030',
'TUMOR': '0/1/2:59,2,56:0.029,0.440'
})
def test_variant_to_string(self):
self.assertEqual(
self.variant.__repr__(),
'chr1 14752 . G A . weak_evidence DP=236;FOO GT:AD:AF 0/0:113,4:0.063 0/1:113,4:0.063'
)
def test_variant_get_info(self):
self.assertEqual(self.variant.get_info('DP'), '236')
def test_variant_get_info_not_set(self):
self.assertIs(self.variant.get_info('TLOD'), False)
def test_variant_set_info_with_value(self):
self.variant.set_info(key='TLOD', value='3.14')
self.assertEqual(self.variant.get_info('TLOD'), '3.14')
def test_variant_set_info_with_value_already_set(self):
self.variant.set_info(key='DP', value='42')
self.assertEqual(self.variant.get_info('DP'), '42')
def test_variant_set_info_without_value(self):
self.variant.set_info(key='STR')
self.assertIs(self.variant.get_info('STR'), True)
def test_variant_set_info_without_value_already_set(self):
self.variant.set_info(key='FOO')
self.assertIs(self.variant.get_info('FOO'), True)
def test_parse_vcf(self):
self.maxDiff = None
vcf_header = [
'##fileformat=VCFv4.2',
'##FILTER=<ID=PASS,Description="Site contains at least one allele that passes filters">',
'##FILTER=<ID=clustered_events,Description="Clustered events observed in the tumor">',
'##FILTER=<ID=weak_evidence,Description="Mutation does not meet likelihood threshold">',
'##FORMAT=<ID=AD,Number=R,Type=Integer,Description="Allelic depths for the ref and alt alleles in the order listed">',
'##FORMAT=<ID=AF,Number=A,Type=Float,Description="Allele fractions of alternate alleles in the tumor">',
'##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">',
'##INFO=<ID=DP,Number=1,Type=Integer,Description="Approximate read depth; some reads may have been filtered">',
'##INFO=<ID=STR,Number=0,Type=Flag,Description="Variant is a short tandem repeat">',
'##contig=<ID=chr1,length=248956422>'
]
vcf_columns = [
'\t'.join([
'#CHROM', 'POS', 'ID', 'REF', 'ALT', 'QUAL', 'FILTER', 'INFO',
'FORMAT', 'NORMAL', 'TUMOR'
])
]
vcf_data = [
'\t'.join([
'chr1', '14752', '.', 'G', 'A', '.', 'weak_evidence', 'DP=236',
'GT:AD:AF', '0/0:113,4:0.063', '0/1:113,4:0.063'
]), '\t'.join([
'chr1', '1625272', '.', 'GC', 'G', '.',
'clustered_events;weak_evidence', 'DP=455;STR', 'GT:AD:AF',
'0|0:224,3:0.015', '0|1:225,3:0.015'
]), '\t'.join([
'chr1', '11117039', '.', 'C', 'A', '.', 'PASS', 'DP=264',
'GT:AD:AF', '0/0:128,0:9.577e-03', '0/1:99,33:0.204'
])
]
vcf_file_lines = [
line + '\n' for line in vcf_header + vcf_columns + vcf_data
]
vcf = parse_vcf(stream=vcf_file_lines)
self.assertEqual(vcf.header, vcf_header)
self.assertEqual(vcf.sample_names, ['NORMAL', 'TUMOR'])
self.assertEqual(vcf.variants, [
Variant(chrom='chr1',
pos='14752',
id_='.',
ref='G',
alt='A',
qual='.',
filter_='weak_evidence',
info='DP=236',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:113,4:0.063',
'TUMOR': '0/1:113,4:0.063'
}),
Variant(chrom='chr1',
pos='1625272',
id_='.',
ref='GC',
alt='G',
qual='.',
filter_='clustered_events;weak_evidence',
info='DP=455;STR',
format_='GT:AD:AF',
samples={
'NORMAL': '0|0:224,3:0.015',
'TUMOR': '0|1:225,3:0.015'
}),
Variant(chrom='chr1',
pos='11117039',
id_='.',
ref='C',
alt='A',
qual='.',
filter_='PASS',
info='DP=264',
format_='GT:AD:AF',
samples={
'NORMAL': '0/0:128,0:9.577e-03',
'TUMOR': '0/1:99,33:0.204'
})
])