def test_denovo_loader_avoids_duplicates(
genome_2013,
fixture_dirname,
fake_families,
):
denovo_filename = fixture_dirname(
"denovo_import/variants_VCF_style_dup.tsv")
params = {
"denovo_chrom": "chrom",
"denovo_pos": "pos",
"denovo_ref": "ref",
"denovo_alt": "alt",
"denovo_family_id": "familyId",
"denovo_best_state": "bestState"
}
variants_loader = DenovoLoader(fake_families,
denovo_filename,
genome=genome_2013,
params=params)
vs = variants_loader.full_variants_iterator()
svs = []
fvs = []
for sv, fvs_ in vs:
print(sv, fvs)
svs.append(sv)
for fv in fvs_:
fvs.append(fv)
assert len(svs) == 3
assert len(fvs) == 4
def test_extra_attributes_serialization_deserialization(
fixtures_gpf_instance, fixture_dirname):
families_data = FamiliesLoader.load_simple_families_file(
fixture_dirname("backends/iossifov_extra_attrs.ped"))
loader = DenovoLoader(
families_data, fixture_dirname("backends/iossifov_extra_attrs.tsv"),
fixtures_gpf_instance.get_genome()
)
main_schema = loader.get_attribute("annotation_schema")
extra_attributes = loader.get_attribute("extra_attributes")
serializer = AlleleParquetSerializer(main_schema, extra_attributes)
it = loader.full_variants_iterator()
variant = next(it)[1][0]
print(variant.gt)
summary_blobs = serializer.serialize_summary_data(variant.alleles)
scores_blob = serializer.serialize_scores_data(variant.alleles)
variant_blob = serializer.serialize_family_variant(
variant.alleles, summary_blobs, scores_blob
)
extra_blob = serializer.serialize_extra_attributes(variant)
family = variant.family
fv = serializer.deserialize_family_variant(
variant_blob, family, extra_blob)
assert fv.get_attribute("someAttr")[0] == "asdf"
def test_extra_attributes_loading_with_person_id(
fixtures_gpf_instance, fixture_dirname):
families_loader = FamiliesLoader(
fixture_dirname("backends/denovo-db-person-id.ped"))
families_data = families_loader.load()
params = {
"denovo_chrom": "Chr",
"denovo_pos": "Position",
"denovo_ref": "Ref",
"denovo_alt": "Alt",
"denovo_person_id": "SampleID"
}
loader = DenovoLoader(
families_data, fixture_dirname("backends/denovo-db-person-id.tsv"),
fixtures_gpf_instance.get_genome(),
params=params
)
it = loader.full_variants_iterator()
variants = list(it)
assert len(variants) == 17
family_variants = [v[1][0] for v in variants]
assert family_variants[0].get_attribute("StudyName")[0] == "Turner_2017"
assert family_variants[1].get_attribute("StudyName")[0] == "Turner_2017"
assert family_variants[2].get_attribute("StudyName")[0] == "Turner_2017"
assert family_variants[3].get_attribute("StudyName")[0] == "Lelieveld2016"
for variant in family_variants:
print(variant)
def test_families_genotypes_decorator_broken_x(fixture_dirname, genome_2013):
families_loader = FamiliesLoader(
fixture_dirname("backends/denovo_families.txt"),
**{"ped_file_format": "simple"},
)
families = families_loader.load()
variants_loader = DenovoLoader(
families, fixture_dirname("backends/denovo_X_broken.txt"), genome_2013)
for sv, fvs in variants_loader.full_variants_iterator():
for fv in fvs:
print(fv, fv.genetic_model)
assert fv.genetic_model == GeneticModel.X_broken
def test_denovo_loader(genome_2013, fixture_dirname, fake_families, filename,
params):
denovo_filename = fixture_dirname(f"denovo_import/{filename}")
variants_loader = DenovoLoader(fake_families,
denovo_filename,
genome=genome_2013,
params=params)
vs = list(variants_loader.full_variants_iterator())
print(vs)
def falt_allele(index):
return vs[index][1][0].alt_alleles[0]
fa = falt_allele(0)
print(fa, fa.variant_in_members, fa.inheritance_in_members)
assert fa.inheritance_in_members[2] == Inheritance.denovo
assert fa.inheritance_in_members[4] == Inheritance.denovo
assert fa.inheritance_in_members == [
Inheritance.unknown,
Inheritance.unknown,
Inheritance.denovo,
Inheritance.missing,
Inheritance.denovo,
]
fa = falt_allele(1)
print(fa, fa.variant_in_members, fa.inheritance_in_members)
assert fa.inheritance_in_members[2] == Inheritance.denovo
assert fa.inheritance_in_members == [
Inheritance.unknown,
Inheritance.unknown,
Inheritance.denovo,
Inheritance.missing,
Inheritance.missing,
]
fa = falt_allele(2)
print(fa, fa.variant_in_members, fa.inheritance_in_members)
assert fa.inheritance_in_members[3] == Inheritance.denovo
assert fa.inheritance_in_members == [
Inheritance.unknown,
Inheritance.unknown,
Inheritance.missing,
Inheritance.denovo,
]
fa = falt_allele(3)
print(fa, fa.variant_in_members, fa.inheritance_in_members)
assert fa.inheritance_in_members[0] == Inheritance.denovo
assert fa.inheritance_in_members == [Inheritance.denovo]
fa = falt_allele(4)
print(fa, fa.variant_in_members, fa.inheritance_in_members)
assert fa.inheritance_in_members[0] == Inheritance.denovo
assert fa.inheritance_in_members == [
Inheritance.denovo,
Inheritance.denovo,
]