def test_gt_quality_threshold(threshold, tmpdir):
thres = 1 - 10**(-threshold / 10.0)
out_vcf = str(tmpdir.join("out.vcf"))
priors_vcf = str(tmpdir.join("priors.vcf"))
run_genotype(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio.vcf",
output=out_vcf,
gt_qual_threshold=threshold,
indels=False,
prioroutput=priors_vcf,
)
for path in [out_vcf, priors_vcf]:
for record in VariantFile(path):
for call in record.samples.values():
likelihoods = call["GL"]
genotype = call["GT"]
if genotype == (None, ):
genotype = None
else:
genotype = genotype[0] + genotype[1]
gt = likeliest_genotype(likelihoods[0], likelihoods[1],
likelihoods[2], thres)
# print(likelihoods[0], likelihoods[1], likelihoods[2], gt, genotype, thres)
assert gt == genotype
def test_genotype_likelihoods_given(tmp_path):
outvcf = tmp_path / "output_gl.vcf"
run_genotype(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio_genotype_likelihoods.vcf",
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True, genotype_likelihoods=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
# check if PL likelihoods (that were present before) are deleted
vcf_reader = VariantFile(outvcf)
# print(list(vcf_reader.samples), outvcf)
for record in vcf_reader:
for call in record.samples.values():
PL = call.get("PL", None)
GL = call.get("GL", None)
print("GL:", GL, "PL:", PL)
assert PL == (None, None, None)
assert GL is not None
def test_genotype_log_likelihoods_given(tmp_path):
outvcf = tmp_path / "output_gl_log.vcf"
outpriors = tmp_path / "priors.vcf"
run_genotype(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio_genotype_log_likelihoods.vcf",
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
gt_qual_threshold=0,
prioroutput=outpriors,
)
for outfile in [outvcf, outpriors]:
assert os.path.isfile(outfile)
tables = list(
VcfReader(outfile, phases=True, genotype_likelihoods=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
# check if GL likelihoods were replaced
vcf_reader = VariantFile(outfile)
print(list(vcf_reader.header.samples), outfile)
for record in vcf_reader:
for call in record.samples.values():
GL = call.get("GL", None)
GQ = call.get("GQ", None)
print("GL:", GL, "GQ", GQ)
assert GL != [-1, -1, -1]
assert GQ != 100
def test_bam_without_readgroup():
run_genotype(
phase_input_files=["tests/data/no-readgroup.bam"],
variant_file="tests/data/onevariant.vcf",
output="/dev/null",
ignore_read_groups=True,
)
def test_genotyping_specific_chromosome(chromosome, tmp_path):
outvcf = tmp_path / "output.vcf"
outpriors = tmp_path / "priors.vcf"
run_genotype(
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],
prioroutput=outpriors,
)
for outfile in [outvcf, outpriors]:
assert os.path.isfile(outfile)
tables = list(VcfReader(outfile, genotype_likelihoods=True))
assert len(tables) == 2
for table in tables:
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
index = 0
if chromosome == "1":
index = 1
# should be no genotype likelihoods for skipped chromosomes
for s in tables[index].samples:
assert tables[index].genotype_likelihoods_of(s) == [None] * 5
assert tables[not index].genotype_likelihoods_of(s) != [None] * 5
def test_ped_sample(sample_set, tmp_path):
# running with --ped and --sample on subset of trio,
# should give same results as running with only --sample
# the trio information should be ignored
outvcf1 = tmp_path / "output1.vcf"
outvcf2 = tmp_path / "output2.vcf"
run_genotype(
phase_input_files=[ped_samples_bamfile],
variant_file="tests/data/ped_samples.vcf",
output=outvcf1,
ped="tests/data/trio.ped",
samples=sample_set,
)
run_genotype(
phase_input_files=[ped_samples_bamfile],
variant_file="tests/data/ped_samples.vcf",
output=outvcf2,
samples=sample_set,
)
assert os.path.isfile(outvcf1)
assert os.path.isfile(outvcf2)
tables1 = list(VcfReader(outvcf1, phases=True, genotype_likelihoods=True))
tables2 = list(VcfReader(outvcf2, phases=True, genotype_likelihoods=True))
assert (len(tables1) == 1) and (len(tables2) == 1)
table1, table2 = tables1[0], tables2[0]
for individual in sample_set:
for var1, var2 in zip(
table1.genotype_likelihoods_of(individual),
table2.genotype_likelihoods_of(individual),
):
print(var1, var2)
assert var1.log10_probs() == var2.log10_probs()
def test_genotyping_one_of_three_individuals(tmp_path):
outvcf = tmp_path / "output.vcf"
outpriors = tmp_path / "priors.vcf"
run_genotype(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio.vcf",
output=outvcf,
samples=["HG003"],
prioroutput=outpriors,
)
for outfile in [outvcf, outpriors]:
assert os.path.isfile(outfile)
tables = list(
VcfReader(outfile, phases=True, genotype_likelihoods=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
# there should be no genotype predictions for HG003/HG002
default_l = math.log10(1 / 3.0)
for l in [
table.genotype_likelihoods_of("HG002"),
table.genotype_likelihoods_of("HG004"),
]:
for var in l:
for v in var.log10_probs():
assert pytest.approx(default_l) == v
def test_no_indels(tmpdir, priors):
prioroutput = str(tmpdir.join("priors.vcf")) if priors else None
outvcf = str(tmpdir.join("output_gl.vcf"))
run_genotype(
phase_input_files=["tests/data/pacbio/pacbio.bam"],
variant_file="tests/data/pacbio/variants.vcf",
reference="tests/data/pacbio/reference.fasta",
output=outvcf,
indels=False,
nopriors=not priors,
prioroutput=prioroutput,
)
result_vcfs = [outvcf]
if priors:
result_vcfs.append(prioroutput)
# make sure indels not genotyped (also in priors.vcf if computed)
for o_vcf in result_vcfs:
vcf_reader = VariantFile(o_vcf)
default_l = math.log10(1 / 3.0)
for record in vcf_reader:
if len(record.alts[0]) != len(record.ref):
for call in record.samples.values():
for v in call["GL"]:
assert pytest.approx(default_l) == v
def test_requested_sample_not_found():
with pytest.raises(CommandLineError):
run_genotype(
phase_input_files=["tests/data/oneread.bam"],
variant_file="tests/data/onevariant.vcf",
output="/dev/null",
samples=["DOES_NOT_EXIST"],
)
def test_wrong_chromosome(tmp_path):
outvcf = tmp_path / "output.vcf"
with pytest.raises(CommandLineError):
run_genotype(
phase_input_files=[short_bamfile],
ignore_read_groups=True,
variant_file="tests/data/short-genome/wrongchromosome.vcf",
output=outvcf,
)
def test_empty_format_field(tmp_path):
outvcf = tmp_path / "output_empty_format.vcf"
run_genotype(
phase_input_files=[trio_bamfile],
variant_file="tests/data/empty_format.vcf",
output=outvcf,
gt_qual_threshold=0,
)
# check if sample fields now contain information
assert os.path.isfile(outvcf)
vcf_reader = VariantFile(outvcf)
for record in vcf_reader:
for sample, call in record.samples.items():
assert set(call) == {"GT", "GL", "GQ"}
def test_multiallelic(tmpdir):
outvcf = str(tmpdir.join("output_multi.vcf"))
run_genotype(
phase_input_files=["tests/data/pacbio/pacbio.bam"],
variant_file="tests/data/multiallelic.vcf",
reference="tests/data/pacbio/reference.fasta",
output=outvcf,
indels=False,
)
vcf_reader = VariantFile(outvcf)
for record in vcf_reader:
n_alleles = len(record.alts) + 1
if n_alleles > 1:
for call in record.samples.values():
assert len(call["GL"]) == ((n_alleles + 1) * n_alleles) / 2
def test_phase_trio_paired_end_reads(tmp_path):
outvcf = tmp_path / "output-paired_end.vcf"
run_genotype(
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"]
def test_adding_constant(constant, tmpdir):
priors_raw_vcf = str(tmpdir.join("output.raw_priors.vcf"))
outvcf_raw_vcf = str(tmpdir.join("output_raw.vcf"))
priors_const_vcf = str(tmpdir.join("output.const_priors.vcf"))
outvcf_const_vcf = str(tmpdir.join("output_raw.vcf"))
# run genotyping without adding constant to priors
run_genotype(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio.vcf",
prioroutput=priors_raw_vcf,
output=outvcf_raw_vcf,
indels=False,
)
# run genotyping with modified priors
run_genotype(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio.vcf",
prioroutput=priors_const_vcf,
output=outvcf_const_vcf,
indels=False,
constant=constant,
)
# check if priors were modified properly
with pysam.VariantFile(priors_raw_vcf) as f:
records_raw = list(f)
with pysam.VariantFile(priors_const_vcf) as f:
records_const = list(f)
assert len(records_raw) == len(records_const)
for record_raw, record_const in zip(records_raw, records_const):
likelihoods_raw = extract_likelihoods(record_raw)
likelihoods_const = extract_likelihoods(record_const)
norm_sum = likelihoods_raw[0] + likelihoods_raw[1] + likelihoods_raw[
2] + 3.0 * constant
# print(float(likelihoods_const[0]), float((likelihoods_raw[0] + constant)/norm_sum))
# print(float(likelihoods_const[1]), float((likelihoods_raw[1] + constant)/norm_sum))
# print(float(likelihoods_const[2]), float((likelihoods_raw[2] + constant)/norm_sum))
for j in range(3):
assert (pytest.approx(likelihoods_const[j],
1e-5) == (likelihoods_raw[j] + constant) /
norm_sum)
def test_genotyping_trio(tmp_path):
outvcf = tmp_path / "output.vcf"
outpriors = tmp_path / "priors.vcf"
run_genotype(
phase_input_files=[trio_bamfile],
variant_file="tests/data/trio.vcf",
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
prioroutput=outpriors,
)
for outfile in [outvcf, outpriors]:
assert os.path.isfile(outfile)
tables = list(VcfReader(outfile, phases=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002"]
def test_use_ped_samples(tmp_path):
outvcf = tmp_path / "output_ped_samples.vcf"
run_genotype(
phase_input_files=[ped_samples_bamfile],
variant_file="tests/data/ped_samples.vcf",
output=outvcf,
ped="tests/data/trio.ped",
genmap="tests/data/trio.map",
use_ped_samples=True,
)
assert os.path.isfile(outvcf)
tables = list(VcfReader(outvcf, phases=True, genotype_likelihoods=True))
assert len(tables) == 1
table = tables[0]
assert table.chromosome == "1"
assert len(table.variants) == 5
assert table.samples == ["HG004", "HG003", "HG002", "orphan"]
default_l = math.log10(1 / 3.0)
for var in table.genotype_likelihoods_of("orphan"):
for v in var.log10_probs():
assert pytest.approx(default_l) == v
def test_one_variant():
run_genotype(
phase_input_files=["tests/data/oneread.bam"],
variant_file="tests/data/onevariant.vcf",
output="/dev/null",
)
def test_with_reference():
run_genotype(
phase_input_files=["tests/data/pacbio/pacbio.bam"],
variant_file="tests/data/pacbio/variants.vcf",
reference="tests/data/pacbio/reference.fasta",
)
def test_default_output():
"""Output to stdout"""
run_genotype(
phase_input_files=["tests/data/oneread.bam"],
variant_file="tests/data/onevariant.vcf",
)
