def test_phase_three_individuals(algorithm, tmpdir):
outvcf = str(tmpdir.join("output.vcf"))
outreadlist = str(tmpdir.join("readlist.tsv"))
run_whatshap(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio.vcf",
read_list_filename=outreadlist,
output=outvcf,
algorithm=algorithm,
)
assert os.path.isfile(outvcf)
assert os.path.isfile(outreadlist)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
phase1 = VariantCallPhase(60906167, (0, 1), None)
phase3 = VariantCallPhase(60907394, (0, 1), None)
assert_phasing(table.phases_of("HG004"),
[None, phase3, phase3, phase3, None])
assert_phasing(table.phases_of("HG003"),
[phase1, None, phase1, None, None])
assert_phasing(table.phases_of("HG002"), [None, None, None, None, None])
def test_phase_trio_use_ped_samples():
with TemporaryDirectory() as tempdir:
for ped_samples in [True, False]:
outvcf = tempdir + '/output_ped_samples.vcf'
outreadlist = tempdir + '/readlist.tsv'
run_whatshap(phase_input_files=[ped_samples_bamfile], variant_file='tests/data/ped_samples.vcf', read_list_filename=outreadlist, output=outvcf,
ped='tests/data/trio.ped', genmap='tests/data/trio.map', use_ped_samples=ped_samples)
assert os.path.isfile(outvcf)
assert os.path.isfile(outreadlist)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 5
assert table.samples == ['HG004', 'HG003', 'HG002', 'orphan']
phase0 = VariantCallPhase(60906167, 0, None)
phase1 = VariantCallPhase(60907394, 0, None)
assert_phasing(table.phases_of('HG004'), [phase0, phase0, phase0, phase0, phase0])
assert_phasing(table.phases_of('HG003'), [phase0, None, phase0, phase0, phase0])
assert_phasing(table.phases_of('HG002'), [None, phase0, None, None, None])
if ped_samples:
assert_phasing(table.phases_of('orphan'), [None, None, None, None, None])
else:
assert_phasing(table.phases_of('orphan'), [None, phase1, phase1, phase1, None])
def test_phase_specific_chromosome():
for requested_chromosome in ['1','2']:
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output.vcf'
run_whatshap(phase_input_files=[trio_bamfile], variant_file='tests/data/trio-two-chromosomes.vcf', output=outvcf,
ped='tests/data/trio.ped', genmap='tests/data/trio.map', chromosomes=[requested_chromosome])
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 2
for table in tables:
assert len(table.variants) == 5
assert table.samples == ['HG004', 'HG003', 'HG002']
if table.chromosome == '1' == requested_chromosome:
phase0 = VariantCallPhase(60906167, 0, None)
assert_phasing(table.phases_of('HG004'), [phase0, phase0, phase0, phase0, phase0])
assert_phasing(table.phases_of('HG003'), [phase0, None, phase0, phase0, phase0])
assert_phasing(table.phases_of('HG002'), [None, phase0, None, None, None])
elif table.chromosome == '2' == requested_chromosome:
phase0 = VariantCallPhase(60906167, 0, None)
phase1 = VariantCallPhase(60906167, 1, None)
assert_phasing(table.phases_of('HG004'), [phase0, None, None, None, phase1])
assert_phasing(table.phases_of('HG003'), [phase0, None, None, None, None])
assert_phasing(table.phases_of('HG002'), [None, None, None, None, phase0])
else:
assert_phasing(table.phases_of('HG004'), [None, None, None, None, None])
assert_phasing(table.phases_of('HG003'), [None, None, None, None, None])
assert_phasing(table.phases_of('HG002'), [None, None, None, None, None])
def test_phase_trio_dont_merge_blocks(tmpdir):
outvcf = str(tmpdir.join("output-merged-blocks.vcf"))
run_whatshap(
phase_input_files=[trio_merged_bamfile],
variant_file="tests/data/trio-merged-blocks.vcf",
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
genetic_haplotyping=False,
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 8
assert table.samples == ["HG002", "HG003", "HG004"]
assert table.num_of_blocks_of("HG004") == 2
assert table.num_of_blocks_of("HG003") == 1
assert table.num_of_blocks_of("HG002") == 1
phase1 = VariantCallPhase(752566, (1, 0), None)
phase2_0 = VariantCallPhase(853954, (0, 1), None)
phase2_1 = VariantCallPhase(853954, (1, 0), None)
assert_phasing(
table.phases_of("HG004"),
[phase1, phase1, phase1, None, phase2_1, phase2_1, phase2_1, phase2_1],
)
assert_phasing(
table.phases_of("HG003"),
[None, None, None, None, phase2_0, phase2_0, phase2_0, phase2_1],
)
assert_phasing(table.phases_of("HG002"),
[None, None, None, None, None, None, None, phase2_1])
def test_phase_trio_paired_end_reads(tmp_path):
outvcf = tmp_path / "output-paired_end.vcf"
run_whatshap(
phase_input_files=[trio_paired_end_bamfile],
variant_file="tests/data/paired_end.sorted.vcf",
output=outvcf,
ped="tests/data/trio_paired_end.ped",
genmap="tests/data/trio.map",
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 3
assert table.samples == ["mother", "father", "child"]
assert table.num_of_blocks_of("mother") == 1
assert table.num_of_blocks_of("father") == 0
assert table.num_of_blocks_of("child") == 1
phase0 = VariantCallPhase(80050, (0, 1), None)
phase1 = VariantCallPhase(80050, (1, 0), None)
assert_phasing(table.phases_of("mother"), [phase1, phase1, phase0])
assert_phasing(table.phases_of("father"), [None, None, None])
assert_phasing(table.phases_of("child"), [None, None, phase1])
def test_phase_three_individuals(algorithm):
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output.vcf'
outreadlist = tempdir + '/readlist.tsv'
run_whatshap(
phase_input_files=[trio_bamfile],
variant_file='tests/data/trio.vcf',
read_list_filename=outreadlist,
output=outvcf,
algorithm=algorithm)
assert os.path.isfile(outvcf)
assert os.path.isfile(outreadlist)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 5
assert table.samples == ['HG004', 'HG003', 'HG002']
phase1 = VariantCallPhase(60906167, 0, None)
phase3 = VariantCallPhase(60907394, 0, None)
assert_phasing(table.phases_of('HG004'), [None, phase3, phase3, phase3, None])
assert_phasing(table.phases_of('HG003'), [phase1, None, phase1, None, None])
assert_phasing(table.phases_of('HG002'), [None, None, None, None, None])
def test_phase_trio_distrust_genotypes(tmpdir):
outvcf = str(tmpdir.join("output_gl.vcf"))
outreadlist = str(tmpdir.join("readlist.tsv"))
run_whatshap(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio_genotype_likelihoods.vcf",
read_list_filename=outreadlist,
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
distrust_genotypes=True,
)
assert os.path.isfile(outvcf)
assert os.path.isfile(outreadlist)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
phase0 = VariantCallPhase(60906167, (0, 1), None)
assert_phasing(table.phases_of("HG004"),
[None, phase0, phase0, phase0, None])
assert_phasing(table.phases_of("HG003"),
[phase0, None, phase0, phase0, phase0])
assert_phasing(table.phases_of("HG002"),
[phase0, None, phase0, phase0, phase0])
def test_phase_quartet_recombination_breakpoints(expect_recombination,
parameters, tmp_path):
outvcf = tmp_path / "output-recombination_breaks.vcf"
outlist = tmp_path / "output.recomb"
run_whatshap(
phase_input_files=[recombination_breaks_bamfile],
variant_file="tests/data/quartet.vcf.gz",
output=outvcf,
ped="tests/data/recombination_breaks.ped",
recombination_list_filename=outlist,
**parameters,
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 4
assert table.samples == ["HG002", "HG005", "HG003", "HG004"]
assert table.num_of_blocks_of("HG002") == 0
assert table.num_of_blocks_of("HG005") == 0
assert table.num_of_blocks_of("HG003") == 1
assert table.num_of_blocks_of("HG004") == 0
phase0 = VariantCallPhase(68735304, (0, 1), None)
phase1 = VariantCallPhase(68735304, (1, 0), None)
assert_phasing(table.phases_of("HG002"), [None, None, None, None])
assert_phasing(table.phases_of("HG005"), [None, None, None, None])
if expect_recombination:
assert_phasing(table.phases_of("HG003"),
[phase0, phase0, None, phase1])
else:
assert_phasing(table.phases_of("HG003"),
[phase0, phase0, None, phase0])
assert_phasing(table.phases_of("HG004"), [None, None, None, None])
lines = open(outlist).readlines()
if expect_recombination:
assert len(lines) == 3
assert lines[1] == "HG002 1 68735433 68738308 0 1 0 0 3\n"
assert lines[2] == "HG005 1 68735433 68738308 0 1 0 0 3\n"
else:
assert len(lines) == 1
def test_read_tetraploid_phased():
tables = list(VcfReader("tests/data/polyploid.chr22.phased.vcf", phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "chr22"
assert table.samples == ["HG00514_NA19240"]
assert len(table.variants) == 8
expected_phase = [
VariantCallPhase(block_id=20000000, phase=(1, 0, 1, 1), quality=None),
VariantCallPhase(block_id=20000000, phase=(1, 0, 1, 0), quality=None),
None,
VariantCallPhase(block_id=20000000, phase=(1, 0, 1, 1), quality=None),
VariantCallPhase(block_id=20001000, phase=(1, 0, 1, 1), quality=None),
VariantCallPhase(block_id=20001000, phase=(0, 0, 0, 1), quality=None),
VariantCallPhase(block_id=20001000, phase=(0, 0, 0, 1), quality=None),
VariantCallPhase(block_id=20001000, phase=(0, 0, 0, 1), quality=None),
]
print("Got:")
for variant in table.phases[0]:
print(variant)
print("Exp:")
for variant in expected_phase:
print(variant)
assert list(table.phases[0]) == expected_phase
def test_phase_quartet_recombination_breakpoints():
parameter_sets = [
(False, {'genmap':'tests/data/recombination_breaks.map'}),
(True, {'recombrate':1000000}),
(False, {'recombrate':.0000001})
]
for expect_recombination, parameters in parameter_sets:
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output-recombination_breaks.vcf'
outlist = tempdir + '/output.recomb'
run_whatshap(phase_input_files=[recombination_breaks_bamfile], variant_file='tests/data/quartet.vcf.gz', output=outvcf,
ped='tests/data/recombination_breaks.ped', recombination_list_filename = outlist, **parameters)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 4
assert table.samples == ['HG002', 'HG005', 'HG003', 'HG004']
assert table.num_of_blocks_of('HG002') == 0
assert table.num_of_blocks_of('HG005') == 0
assert table.num_of_blocks_of('HG003') == 1
assert table.num_of_blocks_of('HG004') == 0
phase0 = VariantCallPhase(68735304, 0, None)
phase1 = VariantCallPhase(68735304, 1, None)
assert_phasing(table.phases_of('HG002'), [None, None, None, None])
assert_phasing(table.phases_of('HG005'), [None, None, None, None])
if expect_recombination:
assert_phasing(table.phases_of('HG003'), [phase0, phase0, None, phase1])
else:
assert_phasing(table.phases_of('HG003'), [phase0, phase0, None, phase0])
assert_phasing(table.phases_of('HG004'), [None, None, None, None])
lines = open(outlist).readlines()
if expect_recombination:
assert len(lines) == 3
assert lines[1]=='HG002 1 68735433 68738308 0 0 0 1 3\n'
assert lines[2]=='HG005 1 68735433 68738308 0 0 0 1 3\n'
else:
assert len(lines) == 1
def test_phase_specific_chromosome(chromosome, tmp_path):
outvcf = tmp_path / "output.vcf"
run_whatshap(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio-two-chromosomes.vcf",
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
chromosomes=[chromosome],
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 2
for table in tables:
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
if table.chromosome == "1" == chromosome:
phase0 = VariantCallPhase(60906167, (0, 1), None)
assert_phasing(
table.phases_of("HG004"),
[phase0, phase0, phase0, phase0, phase0],
)
assert_phasing(table.phases_of("HG003"),
[phase0, None, phase0, phase0, phase0])
assert_phasing(table.phases_of("HG002"),
[None, phase0, None, None, None])
elif table.chromosome == "2" == chromosome:
phase0 = VariantCallPhase(60906167, (0, 1), None)
phase1 = VariantCallPhase(60906167, (1, 0), None)
assert_phasing(table.phases_of("HG004"),
[phase0, None, None, None, phase1])
assert_phasing(table.phases_of("HG003"),
[phase0, None, None, None, None])
assert_phasing(table.phases_of("HG002"),
[None, None, None, None, phase0])
else:
assert_phasing(table.phases_of("HG004"),
[None, None, None, None, None])
assert_phasing(table.phases_of("HG003"),
[None, None, None, None, None])
assert_phasing(table.phases_of("HG002"),
[None, None, None, None, None])
def test_genetic_haplotyping(tmp_path):
outvcf = tmp_path / "output.vcf"
outrecomb = tmp_path / "utput.recomb"
run_whatshap(
variant_file="tests/data/genetic-haplotyping.vcf",
phase_input_files=[],
ped="tests/data/genetic-haplotyping.ped",
output=outvcf,
recombination_list_filename=outrecomb,
)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 3
assert table.samples == [
"sampleA", "sampleB", "sampleC", "sampleD", "sampleE"
]
assert table.num_of_blocks_of("sampleA") == 1
assert table.num_of_blocks_of("sampleB") == 1
assert table.num_of_blocks_of("sampleC") == 0
assert table.num_of_blocks_of("sampleD") == 1
assert table.num_of_blocks_of("sampleE") == 1
phase0 = VariantCallPhase(10327, (0, 1), None)
phase1 = VariantCallPhase(10327, (1, 0), None)
assert_phasing(table.phases_of("sampleA"), [phase0, phase0, phase1])
assert_phasing(table.phases_of("sampleB"), [phase0, None, None])
assert_phasing(table.phases_of("sampleC"), [None, None, None])
assert_phasing(table.phases_of("sampleD"), [phase0, None, phase1])
assert_phasing(table.phases_of("sampleE"), [phase0, phase0, None])
lines = [l.split() for l in open(outrecomb)]
assert len(lines) == 2
Fields = namedtuple("Fields", [f.strip("#\n") for f in lines[0]])
recomb = Fields(*lines[1])
print(recomb)
assert recomb.child_id == "sampleC"
assert recomb.chromosome == "1"
assert recomb.position1 == "31295"
assert recomb.position2 == "102596"
def test_indel_phasing(algorithm):
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output.vcf'
run_whatshap(
phase_input_files=[indels_bamfile],
indels=True, variant_file='tests/data/indels.vcf',
reference='tests/data/random0.fasta',
output=outvcf,
algorithm=algorithm)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, indels=True, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == 'random0'
assert len(table.variants) == 4
assert table.samples == ['sample1']
phase0 = VariantCallPhase(41, 0, None)
phase1 = VariantCallPhase(41, 1, None)
assert_phasing(table.phases_of('sample1'), [phase0, phase1, phase0, phase1])
def test_read_phased_vcf():
for filename in ["tests/data/phased-via-HP.vcf", "tests/data/phased-via-PS.vcf"]:
print("Testing", filename)
tables = list(VcfReader(filename, phases=True))
assert len(tables) == 2
table_a, table_b = tables
assert table_a.chromosome == "chrA"
assert len(table_a.variants) == 4
assert table_a.samples == ["sample1", "sample2"]
assert table_b.chromosome == "chrB"
assert len(table_b.variants) == 2
assert table_b.samples == ["sample1", "sample2"]
assert len(table_a.genotypes) == 2
assert list(table_a.genotypes[0]) == canonic_index_list_to_biallelic_gt_list([1, 2, 1, 1])
assert list(table_a.genotypes[1]) == canonic_index_list_to_biallelic_gt_list([1, 1, 1, 1])
assert list(table_a.genotypes_of("sample1")) == canonic_index_list_to_biallelic_gt_list(
[1, 2, 1, 1]
)
assert list(table_a.genotypes_of("sample2")) == canonic_index_list_to_biallelic_gt_list(
[1, 1, 1, 1]
)
assert len(table_b.genotypes) == 2
assert list(table_b.genotypes[0]) == canonic_index_list_to_biallelic_gt_list([0, 1])
assert list(table_b.genotypes[1]) == canonic_index_list_to_biallelic_gt_list([1, 2])
assert list(table_b.genotypes_of("sample1")) == canonic_index_list_to_biallelic_gt_list(
[0, 1]
)
assert list(table_b.genotypes_of("sample2")) == canonic_index_list_to_biallelic_gt_list(
[1, 2]
)
print(table_a.phases)
assert len(table_a.phases) == 2
expected_phase_sample1 = [
None,
None,
VariantCallPhase(block_id=300, phase=(1, 0), quality=23),
VariantCallPhase(block_id=300, phase=(0, 1), quality=42),
]
expected_phase_sample2 = [
VariantCallPhase(block_id=100, phase=(0, 1), quality=10),
VariantCallPhase(block_id=100, phase=(1, 0), quality=20),
VariantCallPhase(block_id=300, phase=(0, 1), quality=30),
VariantCallPhase(block_id=300, phase=(0, 1), quality=None),
]
assert list(table_a.phases[0]) == expected_phase_sample1
assert list(table_a.phases[1]) == expected_phase_sample2
assert list(table_a.phases_of("sample1")) == expected_phase_sample1
assert list(table_a.phases_of("sample2")) == expected_phase_sample2
assert len(table_b.phases) == 2
assert list(table_b.phases[0]) == [None, None]
assert list(table_b.phases[1]) == [None, None]
assert list(table_b.phases_of("sample1")) == [None, None]
assert list(table_b.phases_of("sample2")) == [None, None]
def test_phase_trio_merged_blocks():
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output-merged-blocks.vcf'
run_whatshap(phase_input_files=[trio_merged_bamfile], variant_file='tests/data/trio-merged-blocks.vcf', output=outvcf,
ped='tests/data/trio.ped', genmap='tests/data/trio.map')
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 8
assert table.samples == ['HG002', 'HG003', 'HG004']
assert table.num_of_blocks_of('HG004') == 1
assert table.num_of_blocks_of('HG003') == 1
assert table.num_of_blocks_of('HG002') == 1
phase0 = VariantCallPhase(752566, 0, None)
phase1 = VariantCallPhase(752566, 1, None)
assert_phasing(table.phases_of('HG004'), [phase1, phase1, phase1, None, phase1, phase1, phase1, phase1])
assert_phasing(table.phases_of('HG003'), [None, None, None, None, phase0, phase0, phase0, phase1])
assert_phasing(table.phases_of('HG002'), [None, None, None, None, None, None, None, phase1])
def test_genetic_haplotyping():
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output.vcf'
outrecomb = tempdir + '/output.recomb'
run_whatshap(variant_file='tests/data/genetic-haplotyping.vcf', phase_input_files=[],
ped='tests/data/genetic-haplotyping.ped', output=outvcf,
recombination_list_filename=outrecomb)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 3
assert table.samples == ['sampleA', 'sampleB', 'sampleC', 'sampleD', 'sampleE']
assert table.num_of_blocks_of('sampleA') == 1
assert table.num_of_blocks_of('sampleB') == 1
assert table.num_of_blocks_of('sampleC') == 0
assert table.num_of_blocks_of('sampleD') == 1
assert table.num_of_blocks_of('sampleE') == 1
phase0 = VariantCallPhase(10327, 0, None)
phase1 = VariantCallPhase(10327, 1, None)
assert_phasing(table.phases_of('sampleA'), [phase0, phase0, phase1])
assert_phasing(table.phases_of('sampleB'), [phase0, None, None])
assert_phasing(table.phases_of('sampleC'), [None, None, None])
assert_phasing(table.phases_of('sampleD'), [phase0, None, phase1])
assert_phasing(table.phases_of('sampleE'), [phase0, phase0, None])
lines = [l.split() for l in open(outrecomb)]
assert len(lines) == 2
Fields = namedtuple('Fields', [f.strip('#\n') for f in lines[0]])
recomb = Fields(*lines[1])
print(recomb)
assert recomb.child_id == 'sampleC'
assert recomb.chromosome == '1'
assert recomb.position1 == '31295'
assert recomb.position2 == '102596'
def test_indel_phasing(algorithm, tmp_path):
outvcf = tmp_path / "output.vcf"
run_whatshap(
phase_input_files=[indels_bamfile],
indels=True,
variant_file="tests/data/indels.vcf",
reference="tests/data/random0.fasta",
output=outvcf,
algorithm=algorithm,
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, indels=True, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "random0"
assert len(table.variants) == 4
assert table.samples == ["sample1"]
phase0 = VariantCallPhase(41, (0, 1), None)
phase1 = VariantCallPhase(41, (1, 0), None)
assert_phasing(table.phases_of("sample1"),
[phase0, phase1, phase0, phase1])
def test_phase_trio_paired_end_reads():
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output-paired_end.vcf'
run_whatshap(phase_input_files=[trio_paired_end_bamfile], variant_file='tests/data/paired_end.sorted.vcf', output=outvcf,
ped='tests/data/trio_paired_end.ped', genmap='tests/data/trio.map')
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 3
assert table.samples == ['mother', 'father', 'child']
assert table.num_of_blocks_of('mother') == 1
assert table.num_of_blocks_of('father') == 0
assert table.num_of_blocks_of('child') == 1
phase0 = VariantCallPhase(80050, 0, None)
phase1 = VariantCallPhase(80050, 1, None)
assert_phasing(table.phases_of('mother'), [phase1, phase1, phase0])
assert_phasing(table.phases_of('father'), [None, None, None])
assert_phasing(table.phases_of('child'), [None, None, phase1])
def test_phase_trio_use_ped_samples(ped_samples, tmpdir):
outvcf = str(tmpdir.join("output_ped_samples.vcf"))
outreadlist = str(tmpdir.join("readlist.tsv"))
run_whatshap(
phase_input_files=[ped_samples_bamfile],
variant_file="tests/data/ped_samples.vcf",
read_list_filename=outreadlist,
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
use_ped_samples=ped_samples,
)
assert os.path.isfile(outvcf)
assert os.path.isfile(outreadlist)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002", "orphan"]
phase0 = VariantCallPhase(60906167, (0, 1), None)
phase1 = VariantCallPhase(60907394, (0, 1), None)
assert_phasing(table.phases_of("HG004"),
[phase0, phase0, phase0, phase0, phase0])
assert_phasing(table.phases_of("HG003"),
[phase0, None, phase0, phase0, phase0])
assert_phasing(table.phases_of("HG002"), [None, phase0, None, None, None])
if ped_samples:
assert_phasing(table.phases_of("orphan"),
[None, None, None, None, None])
else:
assert_phasing(table.phases_of("orphan"),
[None, phase1, phase1, phase1, None])
def test_distrust_genotypes_assertion(tmp_path):
outvcf = tmp_path / "output.vcf"
run_whatshap(
indels=False,
phase_input_files=[dist_geno_bamfile],
variant_file="tests/data/test_dist_geno.vcf",
output=outvcf,
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "chr1"
phase0 = VariantCallPhase(23824647, (0, 1), None)
assert_phasing(table.phases_of("NA12878"), [None, phase0, None, phase0])
def test_phase_one_of_three_individuals():
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output.vcf'
run_whatshap(phase_input_files=[trio_bamfile], variant_file='tests/data/trio.vcf', output=outvcf, samples=['HG003'])
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 5
assert table.samples == ['HG004', 'HG003', 'HG002']
phase0 = VariantCallPhase(60906167,0,None)
assert_phasing(table.phases_of('HG004'), [None, None, None, None, None])
assert_phasing(table.phases_of('HG003'), [phase0, None, phase0, None, None])
assert_phasing(table.phases_of('HG002'), [None, None, None, None, None])
def test_read_phased_vcf():
for filename in [
'tests/data/phased-via-HP.vcf', 'tests/data/phased-via-PS.vcf'
]:
print('Testing', filename)
tables = list(VcfReader(filename, phases=True))
assert len(tables) == 2
table_a, table_b = tables
assert table_a.chromosome == 'chrA'
assert len(table_a.variants) == 4
assert table_a.samples == ['sample1', 'sample2']
assert table_b.chromosome == 'chrB'
assert len(table_b.variants) == 2
assert table_b.samples == ['sample1', 'sample2']
assert len(table_a.genotypes) == 2
assert list(table_a.genotypes[0]) == [1, 2, 1, 1]
assert list(table_a.genotypes[1]) == [1, 1, 1, 1]
assert list(table_a.genotypes_of('sample1')) == [1, 2, 1, 1]
assert list(table_a.genotypes_of('sample2')) == [1, 1, 1, 1]
assert len(table_b.genotypes) == 2
assert list(table_b.genotypes[0]) == [0, 1]
assert list(table_b.genotypes[1]) == [1, 2]
assert list(table_b.genotypes_of('sample1')) == [0, 1]
assert list(table_b.genotypes_of('sample2')) == [1, 2]
print(table_a.phases)
assert len(table_a.phases) == 2
expected_phase_sample1 = [
None, None,
VariantCallPhase(block_id=300, phase=1, quality=23),
VariantCallPhase(block_id=300, phase=0, quality=42)
]
expected_phase_sample2 = [
VariantCallPhase(block_id=100, phase=0, quality=10),
VariantCallPhase(block_id=100, phase=1, quality=20),
VariantCallPhase(block_id=300, phase=0, quality=30),
VariantCallPhase(block_id=300, phase=0, quality=None)
]
assert list(table_a.phases[0]) == expected_phase_sample1
assert list(table_a.phases[1]) == expected_phase_sample2
assert list(table_a.phases_of('sample1')) == expected_phase_sample1
assert list(table_a.phases_of('sample2')) == expected_phase_sample2
assert len(table_b.phases) == 2
assert list(table_b.phases[0]) == [None, None]
assert list(table_b.phases[1]) == [None, None]
assert list(table_b.phases_of('sample1')) == [None, None]
assert list(table_b.phases_of('sample2')) == [None, None]
def test_genetic_phasing_symbolic_alt():
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output.vcf'
run_whatshap(phase_input_files=[], variant_file='tests/data/trio-symbolic-alt.vcf', output=outvcf,
ped='tests/data/trio.ped', indels=True)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True, indels=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 5
assert table.samples == ['HG004', 'HG003', 'HG002']
phase0 = VariantCallPhase(60906167, 0, None)
assert_phasing(table.phases_of('HG004'), [phase0, phase0, phase0, phase0, phase0])
assert_phasing(table.phases_of('HG003'), [phase0, None, phase0, phase0, phase0])
assert_phasing(table.phases_of('HG002'), [None, phase0, None, None, None])
def test_phase_mendelian_conflict():
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output.vcf'
run_whatshap(phase_input_files=[trio_bamfile], variant_file='tests/data/trio-mendelian-conflict.vcf', output=outvcf,
ped='tests/data/trio.ped', genmap='tests/data/trio.map')
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 5
assert table.samples == ['HG004', 'HG003', 'HG002']
phase = VariantCallPhase(60906167, 0, None)
assert_phasing(table.phases_of('HG004'), [phase, None, phase, phase, phase])
assert_phasing(table.phases_of('HG003'), [phase, None, phase, phase, phase])
assert_phasing(table.phases_of('HG002'), [None, None, None, None, None])
def test_phase_trio_distrust_genotypes():
with TemporaryDirectory() as tempdir:
outvcf = tempdir + '/output_gl.vcf'
outreadlist = tempdir + '/readlist.tsv'
run_whatshap(phase_input_files=[trio_bamfile], variant_file='tests/data/trio_genotype_likelihoods.vcf', read_list_filename=outreadlist, output=outvcf,
ped='tests/data/trio.ped', genmap='tests/data/trio.map', distrust_genotypes=True)
assert os.path.isfile(outvcf)
assert os.path.isfile(outreadlist)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == '1'
assert len(table.variants) == 5
assert table.samples == ['HG004', 'HG003', 'HG002']
phase0 = VariantCallPhase(60906167, 0, None)
assert_phasing(table.phases_of('HG004'), [None, phase0, phase0, phase0, None])
assert_phasing(table.phases_of('HG003'), [phase0, None, phase0, phase0, phase0])
assert_phasing(table.phases_of('HG002'), [phase0, None, phase0, phase0, phase0])
def test_phase_missing_genotypes(tmp_path):
outvcf = tmp_path / "output.vcf"
run_whatshap(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio-missing-genotypes.vcf",
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
phase = VariantCallPhase(60906167, (0, 1), None)
assert_phasing(table.phases_of("HG004"), [phase, phase, None, phase, None])
assert_phasing(table.phases_of("HG003"), [phase, None, None, phase, None])
assert_phasing(table.phases_of("HG002"), [None, phase, None, None, None])
def test_phase_mendelian_conflict(tmpdir):
outvcf = str(tmpdir.join("output.vcf"))
run_whatshap(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio-mendelian-conflict.vcf",
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
phase = VariantCallPhase(60906167, (0, 1), None)
assert_phasing(table.phases_of("HG004"),
[phase, None, phase, phase, phase])
assert_phasing(table.phases_of("HG003"),
[phase, None, phase, phase, phase])
assert_phasing(table.phases_of("HG002"), [None, None, None, None, None])
def test_genetic_phasing_symbolic_alt(tmpdir):
outvcf = str(tmpdir.join("output.vcf"))
run_whatshap(
phase_input_files=[],
variant_file="tests/data/trio-symbolic-alt.vcf",
output=outvcf,
ped="tests/data/trio.ped",
indels=True,
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True, indels=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
phase0 = VariantCallPhase(60906167, (0, 1), None)
assert_phasing(table.phases_of("HG004"),
[phase0, phase0, phase0, phase0, phase0])
assert_phasing(table.phases_of("HG003"),
[phase0, None, phase0, phase0, phase0])
assert_phasing(table.phases_of("HG002"), [None, phase0, None, None, None])