def test_process_with_realigner(self):
self.processor.options.mode = deepvariant_pb2.DeepVariantOptions.CALLING
self.processor.options.realigner_options.CopyFrom(
realigner_pb2.RealignerOptions())
self.processor.realigner = mock.Mock()
self.processor.realigner.realign_reads.return_value = [], []
self.processor.sam_reader = mock.Mock()
self.processor.sam_reader.query.return_value = []
self.processor.in_memory_sam_reader = mock.Mock()
c1, c2 = mock.Mock(), mock.Mock()
e1, e2, e3 = mock.Mock(), mock.Mock(), mock.Mock()
self.add_mock('candidates_in_region', retval=([c1, c2], []))
mock_cpe = self.add_mock('create_pileup_examples',
side_effect=[[e1], [e2, e3]])
mock_lv = self.add_mock('label_variant')
self.assertEqual(([c1, c2], [e1, e2, e3], []),
self.processor.process(self.region))
self.processor.sam_reader.query.assert_called_once_with(self.region)
self.processor.realigner.realign_reads.assert_called_once_with(
[], self.region)
self.processor.in_memory_sam_reader.replace_reads.assert_called_once_with(
[])
self.assertEqual([mock.call(c1), mock.call(c2)],
mock_cpe.call_args_list)
test_utils.assert_not_called_workaround(mock_lv)
def realigner_config(flags_obj):
"""Creates a RealignerOptions proto based on input and default settings.
Args:
flags_obj: configuration FLAGS.
Returns:
realigner_pb2.RealignerOptions protobuf.
Raises:
ValueError: If we observe invalid flag values.
"""
ws_config = window_selector_config(flags_obj)
dbg_config = realigner_pb2.DeBruijnGraphOptions(
min_k=flags_obj.dbg_min_k,
max_k=flags_obj.dbg_max_k,
step_k=flags_obj.dbg_step_k,
min_mapq=flags_obj.dbg_min_mapq,
min_base_quality=flags_obj.dbg_min_base_quality,
min_edge_weight=flags_obj.dbg_min_edge_weight,
max_num_paths=flags_obj.dbg_max_num_paths)
aln_config = realigner_pb2.AlignerOptions(
match=flags_obj.aln_match,
mismatch=flags_obj.aln_mismatch,
gap_open=flags_obj.aln_gap_open,
gap_extend=flags_obj.aln_gap_extend,
k=flags_obj.aln_k,
error_rate=flags_obj.aln_error_rate,
max_num_of_mismatches=flags_obj.max_num_mismatches,
realignment_similarity_threshold=flags_obj.
realignment_similarity_threshold,
kmer_size=flags_obj.kmer_size,
force_alignment=False)
diagnostics = realigner_pb2.Diagnostics(
enabled=bool(flags_obj.realigner_diagnostics),
output_root=flags_obj.realigner_diagnostics,
emit_realigned_reads=flags_obj.emit_realigned_reads)
# The normalize_reads flag could came from the `flags_obj` arg, passed in
# from make_examples_options.py. It is already part of AlleleCounterOptions in
# MakeExamplesOptions. Here, we need to set it in RealignerOptions as well
# because an if statement in fast_pass_aligner.cc needs it to decide whether
# to run a specific logic.
# This is not ideal. If there's a way to improve this, please do.
normalize_reads = False
if 'normalize_reads' in flags_obj:
normalize_reads = flags_obj.normalize_reads
return realigner_pb2.RealignerOptions(
ws_config=ws_config,
dbg_config=dbg_config,
aln_config=aln_config,
diagnostics=diagnostics,
split_skip_reads=flags_obj.split_skip_reads,
normalize_reads=normalize_reads)
def realigner_config(flags_obj):
"""Creates a RealignerOptions proto based on input and default settings.
Args:
flags_obj: configuration FLAGS.
Returns:
realigner_pb2.RealignerOptions protobuf.
Raises:
ValueError: If we observe invalid flag values.
"""
ws_config = realigner_pb2.RealignerOptions.WindowSelectorOptions(
min_num_supporting_reads=flags_obj.ws_min_num_supporting_reads,
max_num_supporting_reads=flags_obj.ws_max_num_supporting_reads,
min_mapq=flags_obj.ws_min_mapq,
min_base_quality=flags_obj.ws_min_base_quality,
min_windows_distance=flags_obj.ws_min_windows_distance,
max_window_size=flags_obj.ws_max_window_size)
dbg_config = realigner_pb2.RealignerOptions.DeBruijnGraphOptions(
min_k=flags_obj.dbg_min_k,
max_k=flags_obj.dbg_max_k,
step_k=flags_obj.dbg_step_k,
min_mapq=flags_obj.dbg_min_mapq,
min_base_quality=flags_obj.dbg_min_base_quality,
min_edge_weight=flags_obj.dbg_min_edge_weight,
max_num_paths=flags_obj.dbg_max_num_paths)
aln_config = realigner_pb2.RealignerOptions.AlignerOptions(
match=flags_obj.aln_match,
mismatch=flags_obj.aln_mismatch,
gap_open=flags_obj.aln_gap_open,
gap_extend=flags_obj.aln_gap_extend,
k=flags_obj.aln_k,
error_rate=flags_obj.aln_error_rate)
diagnostics = realigner_pb2.RealignerOptions.Diagnostics(
enabled=bool(flags_obj.realigner_diagnostics),
output_root=flags_obj.realigner_diagnostics,
emit_realigned_reads=flags_obj.emit_realigned_reads)
return realigner_pb2.RealignerOptions(
ws_config=ws_config,
dbg_config=dbg_config,
aln_config=aln_config,
diagnostics=diagnostics)
def realigner_config(flags_obj):
"""Creates a RealignerOptions proto based on input and default settings.
Args:
flags_obj: configuration FLAGS.
Returns:
realigner_pb2.RealignerOptions protobuf.
Raises:
ValueError: If we observe invalid flag values.
"""
ws_config = window_selector_config(flags_obj)
dbg_config = realigner_pb2.DeBruijnGraphOptions(
min_k=flags_obj.dbg_min_k,
max_k=flags_obj.dbg_max_k,
step_k=flags_obj.dbg_step_k,
min_mapq=flags_obj.dbg_min_mapq,
min_base_quality=flags_obj.dbg_min_base_quality,
min_edge_weight=flags_obj.dbg_min_edge_weight,
max_num_paths=flags_obj.dbg_max_num_paths)
aln_config = realigner_pb2.AlignerOptions(
match=flags_obj.aln_match,
mismatch=flags_obj.aln_mismatch,
gap_open=flags_obj.aln_gap_open,
gap_extend=flags_obj.aln_gap_extend,
k=flags_obj.aln_k,
error_rate=flags_obj.aln_error_rate,
max_num_of_mismatches=flags_obj.max_num_mismatches,
realignment_similarity_threshold=flags_obj.
realignment_similarity_threshold,
kmer_size=flags_obj.kmer_size,
force_alignment=False)
diagnostics = realigner_pb2.Diagnostics(
enabled=bool(flags_obj.realigner_diagnostics),
output_root=flags_obj.realigner_diagnostics,
emit_realigned_reads=flags_obj.emit_realigned_reads)
return realigner_pb2.RealignerOptions(ws_config=ws_config,
dbg_config=dbg_config,
aln_config=aln_config,
diagnostics=diagnostics)
def test_process_with_realigner(self):
self.processor.options.mode = deepvariant_pb2.MakeExamplesOptions.CALLING
self.processor.options.realigner_enabled = True
self.processor.options.realigner_options.CopyFrom(
realigner_pb2.RealignerOptions())
self.processor.realigner = mock.Mock()
self.processor.realigner.realign_reads.return_value = [], []
main_sample = self.processor.samples[0]
main_sample.sam_readers = [mock.Mock()]
main_sample.sam_readers[0].query.return_value = []
c1, c2 = mock.Mock(), mock.Mock()
e1, e2, e3 = mock.Mock(), mock.Mock(), mock.Mock()
self.add_mock('candidates_in_region',
retval=({
'main_sample': [c1, c2]
}, {
'main_sample': []
}))
mock_cpe = self.add_mock('create_pileup_examples',
side_effect=[[e1], [e2, e3]])
mock_lc = self.add_mock('label_candidates')
candidates, examples, gvcfs, runtimes = self.processor.process(
self.region)
self.assertEqual(candidates['main_sample'], [c1, c2])
self.assertEqual(examples['main_sample'], [e1, e2, e3])
self.assertEmpty(gvcfs['main_sample'])
self.assertIsInstance(runtimes, dict)
main_sample.sam_readers[0].query.assert_called_once_with(self.region)
self.processor.realigner.realign_reads.assert_called_once_with(
[], self.region)
main_sample.in_memory_sam_reader.replace_reads.assert_called_once_with(
[])
self.assertEqual(
[mock.call(c1, sample_order=[0]),
mock.call(c2, sample_order=[0])], mock_cpe.call_args_list)
test_utils.assert_not_called_workaround(mock_lc)