def __init__(
self,
info_keys_to_move_to_calls_regex, # type: str
copy_quality_to_calls, # type: bool
copy_filter_to_calls, # type: bool
window_size=DEFAULT_WINDOW_SIZE # type: int
):
# type: (...) -> None
"""Initializes the strategy.
Args:
info_keys_to_move_to_calls_regex: A regular expression specifying info
fields that should be moved to calls.
copy_quality_to_calls: Whether to copy the quality field to the associated
calls in each record.
copy_filter_to_calls: Whether to copy filter field to the associated
calls in each record.
window_size: Size of windows that variants will be grouped in based on the
start position of the variant.
"""
self._move_to_calls = move_to_calls_strategy.MoveToCallsStrategy(
info_keys_to_move_to_calls_regex=info_keys_to_move_to_calls_regex,
copy_quality_to_calls=copy_quality_to_calls,
copy_filter_to_calls=copy_filter_to_calls)
self._window_size = window_size
def test_get_merge_keys(self):
strategy = move_to_calls_strategy.MoveToCallsStrategy(None, None, None)
def get_expected_key(reference_name, start, end, reference_bases,
alternate_bases):
return '%s:%s:%s:%s:%s' % (
reference_name or '', str(start or ''), str(
end or ''), strategy._get_hash(reference_bases or ''),
strategy._get_hash(','.join(alternate_bases or [])))
variant = vcfio.Variant()
self.assertEqual(get_expected_key(None, None, None, None, None),
next(strategy.get_merge_keys(variant)))
variant.reference_name = '19'
self.assertEqual(get_expected_key(19, None, None, None, None),
next(strategy.get_merge_keys(variant)))
variant.start = 123
variant.end = 125
variant.reference_bases = 'AT'
self.assertEqual(get_expected_key(19, 123, 125, 'AT', None),
next(strategy.get_merge_keys(variant)))
variant.alternate_bases = ['A', 'C']
self.assertEqual(get_expected_key(19, 123, 125, 'AT', ['A', 'C']),
next(strategy.get_merge_keys(variant)))
def test_get_merged_variants_no_custom_options(self):
strategy = move_to_calls_strategy.MoveToCallsStrategy(
info_keys_to_move_to_calls_regex=None,
copy_quality_to_calls=False,
copy_filter_to_calls=False)
variants = self._get_sample_variants()
# Test single variant merge.
self.assertEqual([variants[0]],
strategy.get_merged_variants([variants[0]]))
# Test multiple variant merge.
merged_variant = strategy.get_merged_variants(variants)[0]
self._assert_common_expected_merged_fields(merged_variant)
self.assertEqual([
vcfio.VariantCall(name='Sample1',
genotype=[0, 1],
info={
'GQ': 20,
'HQ': [10, 20]
}),
vcfio.VariantCall(name='Sample2',
genotype=[1, 0],
info={
'GQ': 10,
'FLAG1': True
}),
vcfio.VariantCall(name='Sample3', genotype=[1, 1]),
vcfio.VariantCall(name='Sample4', genotype=[1, 0], info={'GQ': 20})
], merged_variant.calls)
self.assertItemsEqual(['A1', 'A2', 'A3'], merged_variant.info.keys())
self.assertTrue(merged_variant.info['A1'] in ('some data',
'some data2'))
self.assertEqual(['data1', 'data2'], merged_variant.info['A2'])
self.assertEqual(['data3', 'data4'], merged_variant.info['A3'])
def test_get_merged_variants_move_info_to_calls(self):
strategy = move_to_calls_strategy.MoveToCallsStrategy(
info_keys_to_move_to_calls_regex='^A1$',
copy_quality_to_calls=False,
copy_filter_to_calls=False)
variants = self._get_sample_variants()
# Test single variant merge.
single_merged_variant = strategy.get_merged_variants([variants[0]])[0]
self.assertEqual([
vcfio.VariantCall(sample_id=hash_name('Sample1'),
genotype=[0, 1],
info={
'GQ': 20,
'HQ': [10, 20],
'A1': 'some data'
}),
vcfio.VariantCall(sample_id=hash_name('Sample2'),
genotype=[1, 0],
info={
'GQ': 10,
'FLAG1': True,
'A1': 'some data'
})
], single_merged_variant.calls)
# Test multiple variant merge.
merged_variant = strategy.get_merged_variants(variants)[0]
self._assert_common_expected_merged_fields(merged_variant)
self.assertEqual([
vcfio.VariantCall(sample_id=hash_name('Sample1'),
genotype=[0, 1],
info={
'GQ': 20,
'HQ': [10, 20],
'A1': 'some data'
}),
vcfio.VariantCall(sample_id=hash_name('Sample2'),
genotype=[1, 0],
info={
'GQ': 10,
'FLAG1': True,
'A1': 'some data'
}),
vcfio.VariantCall(sample_id=hash_name('Sample3'),
genotype=[1, 1],
info={'A1': 'some data2'}),
vcfio.VariantCall(sample_id=hash_name('Sample4'),
genotype=[1, 0],
info={
'GQ': 20,
'A1': 'some data2'
})
], merged_variant.calls)
self.assertItemsEqual(['A2', 'A3'], merged_variant.info.keys())
self.assertEqual(['data1', 'data2'], merged_variant.info['A2'])
self.assertEqual(['data3', 'data4'], merged_variant.info['A3'])
def test_modify_bigquery_schema_duplicate_keys(self):
strategy = move_to_calls_strategy.MoveToCallsStrategy(
info_keys_to_move_to_calls_regex='.*',
copy_quality_to_calls=True,
copy_filter_to_calls=True)
info_keys = [ColumnKeyConstants.CALLS_SAMPLE_ID]
base_schema = self._get_base_schema(info_keys)
try:
strategy.modify_bigquery_schema(base_schema, info_keys)
self.fail('Duplicate keys should throw error.')
except ValueError:
pass
def test_modify_bigquery_schema_move_info_to_calls(self):
strategy = move_to_calls_strategy.MoveToCallsStrategy(
info_keys_to_move_to_calls_regex='INFO.*1',
copy_quality_to_calls=False,
copy_filter_to_calls=False)
info_keys = ['INFO_KEY1', 'INFO_KEY2']
base_schema = self._get_base_schema(info_keys)
strategy.modify_bigquery_schema(base_schema, info_keys)
self.assertEqual([
ColumnKeyConstants.REFERENCE_NAME, ColumnKeyConstants.QUALITY,
ColumnKeyConstants.FILTER, ColumnKeyConstants.CALLS, '.'.join([
ColumnKeyConstants.CALLS, ColumnKeyConstants.CALLS_SAMPLE_ID
]), '.'.join([ColumnKeyConstants.CALLS, 'INFO_KEY1']), 'INFO_KEY2'
], self._get_fields_from_schema(base_schema))
def test_merge_variants(self):
variant_merger = move_to_calls_strategy.MoveToCallsStrategy(
'^A1$', False, False)
variant_list, merged_variant = self._get_sample_merged_variants()
unmerged_variant_list = self._get_sample_unmerged_variants()
pipeline = TestPipeline()
merged_variants = (
pipeline
| Create(variant_list + unmerged_variant_list, reshuffle=False)
| 'MergeVariants' >> merge_variants.MergeVariants(variant_merger))
assert_that(merged_variants,
asserts.variants_equal_to_ignore_order([merged_variant] +
unmerged_variant_list))
pipeline.run()
def _get_variant_merge_strategy(known_args # type: argparse.Namespace
):
# type: (...) -> Optional(variant_merge_strategy.VariantMergeStrategy)
merge_options = variant_transform_options.MergeOptions
if (not known_args.variant_merge_strategy
or known_args.variant_merge_strategy == merge_options.NONE):
return None
elif known_args.variant_merge_strategy == merge_options.MOVE_TO_CALLS:
return move_to_calls_strategy.MoveToCallsStrategy(
known_args.info_keys_to_move_to_calls_regex,
known_args.copy_quality_to_calls, known_args.copy_filter_to_calls)
elif (known_args.variant_merge_strategy ==
merge_options.MERGE_WITH_NON_VARIANTS):
return merge_with_non_variants_strategy.MergeWithNonVariantsStrategy(
known_args.info_keys_to_move_to_calls_regex,
known_args.copy_quality_to_calls, known_args.copy_filter_to_calls)
else:
raise ValueError('Merge strategy is not supported.')
def test_get_merged_variants_move_everything_to_calls(self):
strategy = move_to_calls_strategy.MoveToCallsStrategy(
info_keys_to_move_to_calls_regex='.*',
copy_quality_to_calls=True,
copy_filter_to_calls=True)
variants = self._get_sample_variants()
# Test single variant merge.
single_merged_variant = strategy.get_merged_variants([variants[0]])[0]
self.assertEqual([
vcfio.VariantCall(sample_id=hash_name('Sample1'),
genotype=[0, 1],
info={
'GQ': 20,
'HQ': [10, 20],
'A1': 'some data',
'A2': ['data1', 'data2'],
ColumnKeyConstants.QUALITY: 2,
ColumnKeyConstants.FILTER: ['PASS']
}),
vcfio.VariantCall(sample_id=hash_name('Sample2'),
genotype=[1, 0],
info={
'GQ': 10,
'FLAG1': True,
'A1': 'some data',
'A2': ['data1', 'data2'],
ColumnKeyConstants.QUALITY: 2,
ColumnKeyConstants.FILTER: ['PASS']
})
], single_merged_variant.calls)
merged_variant = strategy.get_merged_variants(variants)[0]
self._assert_common_expected_merged_fields(merged_variant)
self.assertEqual([
vcfio.VariantCall(sample_id=hash_name('Sample1'),
genotype=[0, 1],
info={
'GQ': 20,
'HQ': [10, 20],
'A1': 'some data',
'A2': ['data1', 'data2'],
ColumnKeyConstants.QUALITY: 2,
ColumnKeyConstants.FILTER: ['PASS']
}),
vcfio.VariantCall(sample_id=hash_name('Sample2'),
genotype=[1, 0],
info={
'GQ': 10,
'FLAG1': True,
'A1': 'some data',
'A2': ['data1', 'data2'],
ColumnKeyConstants.QUALITY: 2,
ColumnKeyConstants.FILTER: ['PASS']
}),
vcfio.VariantCall(sample_id=hash_name('Sample3'),
genotype=[1, 1],
info={
'A1': 'some data2',
'A3': ['data3', 'data4'],
ColumnKeyConstants.QUALITY: 20,
ColumnKeyConstants.FILTER: ['q10']
}),
vcfio.VariantCall(sample_id=hash_name('Sample4'),
genotype=[1, 0],
info={
'GQ': 20,
'A1': 'some data2',
'A3': ['data3', 'data4'],
ColumnKeyConstants.QUALITY: 20,
ColumnKeyConstants.FILTER: ['q10']
})
], merged_variant.calls)
self.assertEqual([], merged_variant.info.keys())
def test_get_merged_variants_move_quality_and_filter_to_calls(self):
strategy = move_to_calls_strategy.MoveToCallsStrategy(
info_keys_to_move_to_calls_regex='',
copy_quality_to_calls=True,
copy_filter_to_calls=True)
variants = self._get_sample_variants()
# Test single variant merge.
single_merged_variant = strategy.get_merged_variants([variants[0]])[0]
self.assertEqual([
vcfio.VariantCall(name='Sample1',
genotype=[0, 1],
info={
'GQ': 20,
'HQ': [10, 20],
ColumnKeyConstants.QUALITY: 2,
ColumnKeyConstants.FILTER: ['PASS']
}),
vcfio.VariantCall(name='Sample2',
genotype=[1, 0],
info={
'GQ': 10,
'FLAG1': True,
ColumnKeyConstants.QUALITY: 2,
ColumnKeyConstants.FILTER: ['PASS']
})
], single_merged_variant.calls)
# Test multiple variant merge.
merged_variant = strategy.get_merged_variants(variants)[0]
self._assert_common_expected_merged_fields(merged_variant)
self.assertEqual([
vcfio.VariantCall(name='Sample1',
genotype=[0, 1],
info={
'GQ': 20,
'HQ': [10, 20],
ColumnKeyConstants.QUALITY: 2,
ColumnKeyConstants.FILTER: ['PASS']
}),
vcfio.VariantCall(name='Sample2',
genotype=[1, 0],
info={
'GQ': 10,
'FLAG1': True,
ColumnKeyConstants.QUALITY: 2,
ColumnKeyConstants.FILTER: ['PASS']
}),
vcfio.VariantCall(name='Sample3',
genotype=[1, 1],
info={
ColumnKeyConstants.QUALITY: 20,
ColumnKeyConstants.FILTER: ['q10']
}),
vcfio.VariantCall(name='Sample4',
genotype=[1, 0],
info={
'GQ': 20,
ColumnKeyConstants.QUALITY: 20,
ColumnKeyConstants.FILTER: ['q10']
})
], merged_variant.calls)
self.assertItemsEqual(['A1', 'A2', 'A3'], merged_variant.info.keys())
self.assertTrue(merged_variant.info['A1'].data in ('some data',
'some data2'))
self.assertEqual(vcfio.VariantInfo(['data1', 'data2'], '2'),
merged_variant.info['A2'])
self.assertEqual(vcfio.VariantInfo(['data3', 'data4'], '2'),
merged_variant.info['A3'])