def test_read_sample_data(self):
schema = self.__get_example_schema("vcf_example.vcf")
sample_schema = [key for key, _ in schema.iter_sample_data()]
sample_data = SampleData(sample_schema, ['sample1'])
sample_data.add_sample_data("sample1", "GT", GenotypeCall("1|0"))
sample_data.add_sample_data("sample1", "PL", [3000, 0, 3000])
sample_data.add_sample_data("sample1", "GQ", [1000])
sample_data.add_sample_data("sample1", "PQ", [2000])
sample_data.add_sample_data("sample1", "PS", [60000])
sample_data.add_sample_data("sample1", "AD", [140, 110])
sample_data.add_sample_data("sample1", "DP", [250])
sample_data.add_sample_data("sample1", "VAF", [0.4])
self.assertTrue(sample_data.has_sample("sample1"))
self.assertEqual(sample_data.genotypes(),
{"sample1": GenotypeCall("1|0")})
self.assertEqual(sample_data.get_field("sample1", 'GT'),
GenotypeCall("1|0"))
self.assertEqual(sample_data.get_field("sample1", 'PL'),
[3000, 0, 3000])
genotype_data = sample_data.get_genotype_data("sample1")
self.assertEqual(genotype_data.genotype(), GenotypeCall("1|0"))
self.assertEqual(genotype_data['GT'], GenotypeCall("1|0"))
self.assertEqual(genotype_data['PL'], [3000, 0, 3000])
def test_should_merge_genotype_call_object_in_sample_data(self):
sample_data1 = SampleData(['GT'], ['sample_name'])
sample_data1.add_sample_data('sample_name', 'GT', GenotypeCall('0/1'))
sample_data2 = SampleData(['GT'], ['sample_name'])
sample_data2.add_sample_data('sample_name', 'GT', GenotypeCall('0/1'))
sample_data1.merge_genotype_calls(sample_data2.genotypes())
self.assertEqual(sample_data1.get_field("sample_name", "GT"),
GenotypeCall("1/1"))
def test_should_fail_if_sample_data_objects_have_different_sample(self):
sample_data1 = SampleData(['GT'], ['sample_name_1'])
sample_data1.add_sample_data('sample_name_1', 'GT',
GenotypeCall('0/0'))
sample_data2 = SampleData(['GT'], ['sample_name_2'])
sample_data2.add_sample_data('sample_name_2', 'GT',
GenotypeCall('0/0'))
self.assertRaises(Exception, sample_data1.merge_genotype_calls,
sample_data2.genotypes())
def test_should_return_default_diploid_genotype(self):
sample_data = SampleData(['GT', 'GL'], ["NA12878"])
self.assertEqual(GenotypeCall("./."), GenotypeCall("./."))
self.assertTrue(sample_data.has_sample("NA12878"))
self.assertEqual(sample_data.genotypes(),
{"NA12878": GenotypeCall("./.")})
self.assertEqual(sample_data.get_field("NA12878", 'GT'),
GenotypeCall("./."))
self.assertEqual(sample_data.get_field("NA12878", 'GL'), [])
genotype_data = sample_data.get_genotype_data("NA12878")
self.assertEqual(genotype_data.genotype(), GenotypeCall("./."))
self.assertEqual(genotype_data['GT'], GenotypeCall("./."))
self.assertEqual(genotype_data['GL'], [])