def test_exclude_duplicates(self):
""" test that exclude duplicates works correctly
"""
# create a variant that is within two genes
snv1 = create_variant("F", "missense_variant|missense_variant", "TEST1|TEST2")
# two variants that lie in different genes on different chromosomes
# should not be merged
snv2 = create_variant("F", "missense_variant", "OTHER1", chrom="2")
variants = [(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"]),
((snv2, ["single_variant"], ["Monoallelic"], ["OTHER1"]))]
self.assertEqual(sorted(self.finder.exclude_duplicates(variants)), sorted(variants))
# create a list of variant tuples that passed filtering for two
# different gene symbols
variants = [(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"]),
((snv1, ["compound_het"], ["Biallelic"], ["TEST1"])),
((snv1, ["compound_het"], ["Biallelic"], ["TEST1"]))]
self.assertEqual(self.finder.exclude_duplicates(variants),
[(snv1, ["single_variant", "compound_het"], ["Monoallelic", "Biallelic"], ["TEST1"])])
# create a list of variant tuples that passed filtering for two
# different gene symbols
variants = [(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"]),
((snv1, ["single_variant"], ["Monoallelic"], ["TEST2"]))]
# the same variant passing for two gene symbols should be collapsed
# into a single entry, where the entry contains a list ofall the gene
# symbols
self.assertEqual(self.finder.exclude_duplicates(variants),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST1", "TEST2"])])
def test_create_gene_dict(self):
""" test that create_gene_dict works correctly
"""
# create variants that share genes, or not
snv1 = create_variant("F", "missense_variant|missense_variant", "TEST1|TEST2")
snv2 = create_variant("F", "missense_variant", "TEST1")
snv3 = create_variant("F", "missense_variant", "OTHER1")
# the variants that share a gene should be grouped in lists indexed by
# the gene key
self.assertEqual(self.finder.create_gene_dict([snv1, snv2, snv3]),
{"TEST1": [snv1, snv2], "TEST2": [snv1], "OTHER1": [snv3]})
def test_find_variants(self):
""" test that find_variants() works correctly
"""
# define the trio, so that we can know whether the parents are affected.
# The child also needs to be included and set, so that we can get the
# child ID for logging purposes.
family = Family("famID")
family.add_child("child_id", 'dad_id', 'mom_id', 'f', '2', "/vcf/path")
family.add_father("dad_id", '0', '0', 'm', '1', "/vcf/path")
family.add_mother("mom_id", '0', '0', 'f', '1', "/vcf/path")
family.set_child()
# create variants that cover various scenarios
snv1 = create_variant("F", "missense_variant|missense_variant", "TEST1|TEST2")
snv2 = create_variant("F", "missense_variant|synonymous_variant", "OTHER1|OTHER2")
snv3 = create_variant("F", "missense_variant", "")
snv4 = create_variant("F", "missense_variant", "TESTX", chrom="X")
self.finder.known_genes = {"TEST1": {"inh": ["Monoallelic"]},
"OTHER1": {"inh": ["Monoallelic"]},
"OTHER2": {"inh": ["Monoallelic"]},
"TESTX": {"inh": ["X-linked dominant"]}}
# check the simplest case, a variant in a known gene
self.assertEqual(self.finder.find_variants([snv1], "TEST1", family),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"])])
# check that a gene not in a known gene does not pass
self.assertEqual(self.finder.find_variants([snv1], "TEST2", family), [])
# check a variant where the gene is known, but the consequence for that
# gene is not functional, does not pass
self.assertEqual(self.finder.find_variants([snv2], "OTHER2", family), [])
# check that intergenic variants (which lack HGNC symbols) do not pass
self.assertEqual(self.finder.find_variants([snv3], None, family), [])
# check that a variant on chrX passes through the allosomal instance
self.assertEqual(self.finder.find_variants([snv4], "TESTX", family),
[(snv4, ["single_variant"], ["X-linked dominant"], ["TESTX"])])
# remove the known genes, so that the variants in unknown genes pass
self.finder.known_genes = None
self.assertEqual(sorted(self.finder.find_variants([snv1], "TEST2", family)),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST2"]),
(snv1, ["single_variant"], ["Mosaic"], ["TEST2"])])
# but variants without gene symbols still are excluded
self.assertEqual(self.finder.find_variants([snv3], None, family), [])
def test_create_gene_dict(self):
""" test that create_gene_dict works correctly
"""
# create variants that share genes, or not
snv1 = create_variant("F", "missense_variant|missense_variant",
"TEST1|TEST2")
snv2 = create_variant("F", "missense_variant", "TEST1")
snv3 = create_variant("F", "missense_variant", "OTHER1")
# the variants that share a gene should be grouped in lists indexed by
# the gene key
self.assertEqual(self.finder.create_gene_dict([snv1, snv2, snv3]), {
"TEST1": [snv1, snv2],
"TEST2": [snv1],
"OTHER1": [snv3]
})
def test_exclude_duplicates(self):
""" test that exclude duplicates works correctly
"""
# create a variant that is within two genes
snv1 = create_variant("F", "missense_variant|missense_variant",
"TEST1|TEST2")
# two variants that lie in different genes on different chromosomes
# should not be merged
snv2 = create_variant("F", "missense_variant", "OTHER1", chrom="2")
variants = [(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"]),
((snv2, ["single_variant"], ["Monoallelic"], ["OTHER1"]))]
self.assertEqual(sorted(self.finder.exclude_duplicates(variants)),
sorted(variants))
# create a list of variant tuples that passed filtering for two
# different gene symbols
variants = [(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"]),
((snv1, ["compound_het"], ["Biallelic"], ["TEST1"])),
((snv1, ["compound_het"], ["Biallelic"], ["TEST1"]))]
self.assertEqual(self.finder.exclude_duplicates(variants),
[(snv1, ["compound_het", "single_variant"
], ["Biallelic", "Monoallelic"], ["TEST1"])])
# create a list of variant tuples that passed filtering for two
# different gene symbols
variants = [(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"]),
((snv1, ["single_variant"], ["Monoallelic"], ["TEST2"]))]
# the same variant passing for two gene symbols should be collapsed
# into a single entry, where the entry contains a list ofall the gene
# symbols
self.assertEqual(
self.finder.exclude_duplicates(variants),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST1", "TEST2"])])
def test_find_variants(self):
""" test that find_variants() works correctly
"""
# define the trio, so that we can know whether the parents are affected.
# The child also needs to be included and set, so that we can get the
# child ID for logging purposes.
family = Family("famID")
family.add_child("child_id", 'dad_id', 'mom_id', 'f', '2', "/vcf/path")
family.add_father("dad_id", '0', '0', 'm', '1', "/vcf/path")
family.add_mother("mom_id", '0', '0', 'f', '1', "/vcf/path")
family.set_child()
# create variants that cover various scenarios
snv1 = create_variant("F", "missense_variant|missense_variant",
"TEST1|TEST2")
snv2 = create_variant("F", "missense_variant|synonymous_variant",
"OTHER1|OTHER2")
snv3 = create_variant("F", "missense_variant", "")
snv4 = create_variant("F", "missense_variant", "TESTX", chrom="X")
self.finder.known_genes = {
"TEST1": {
"inh": ["Monoallelic"]
},
"OTHER1": {
"inh": ["Monoallelic"]
},
"OTHER2": {
"inh": ["Monoallelic"]
},
"TESTX": {
"inh": ["X-linked dominant"]
}
}
# check the simplest case, a variant in a known gene
self.assertEqual(
self.finder.find_variants([snv1], "TEST1", family),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"])])
# check that a gene not in a known gene does not pass
self.assertEqual(self.finder.find_variants([snv1], "TEST2", family),
[])
# check a variant where the gene is known, but the consequence for that
# gene is not functional, does not pass
self.assertEqual(self.finder.find_variants([snv2], "OTHER2", family),
[])
# check that intergenic variants (which lack HGNC symbols) do not pass
self.assertEqual(self.finder.find_variants([snv3], None, family), [])
# check that a variant on chrX passes through the allosomal instance
self.assertEqual(
self.finder.find_variants([snv4], "TESTX", family),
[(snv4, ["single_variant"], ["X-linked dominant"], ["TESTX"])])
# remove the known genes, so that the variants in unknown genes pass
self.finder.known_genes = None
self.assertEqual(
sorted(self.finder.find_variants([snv1], "TEST2", family)),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST2"]),
(snv1, ["single_variant"], ["Mosaic"], ["TEST2"])])
# but variants without gene symbols still are excluded
self.assertEqual(self.finder.find_variants([snv3], None, family), [])