def testGetNoneShapeFromEmptyExamplesPath(self, file_name_to_write,
tfrecord_path_to_match):
output_file = test_utils.test_tmpfile(file_name_to_write)
io_utils.write_tfrecords([], output_file)
self.assertIsNone(
tf_utils.get_shape_from_examples_path(
test_utils.test_tmpfile(tfrecord_path_to_match)))
def test_make_examples_runtime_by_region(self):
region = ranges.parse_literal('chr20:10,000,000-10,010,000')
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.reads = testdata.CHR20_BAM
FLAGS.regions = [ranges.to_literal(region)]
FLAGS.mode = 'calling'
num_shards = 4
FLAGS.examples = test_utils.test_tmpfile(
_sharded('examples.tfrecord', num_shards))
# Use same number of shards for profiling files as examples.
output_prefix = test_utils.test_tmpfile('runtime_profile')
FLAGS.runtime_by_region = output_prefix + '@{}'.format(num_shards)
FLAGS.task = 2
# Run make_examples with those FLAGS.
options = make_examples.default_options(add_flags=True)
make_examples_core.make_examples_runner(options)
# Sharded output ending in @4 becomes -00002-of-00004 for task 2.
expected_output_path = output_prefix + '-0000{}-of-00004'.format(FLAGS.task)
expected_columns = [
'region', 'get reads', 'find candidates', 'make pileup images',
'write outputs', 'num reads', 'num candidates', 'num examples'
]
with gfile.Open(expected_output_path, 'r') as fin:
header = fin.readline()
column_names = header.strip().split('\t')
self.assertEqual(expected_columns, column_names)
non_header_lines = fin.readlines()
self.assertLen(non_header_lines, 3)
one_row = non_header_lines[0].strip().split('\t')
self.assertEqual(len(one_row), len(column_names))
self.assertGreater(int(one_row[5]), 0, msg='num reads > 0')
self.assertGreater(int(one_row[6]), 0, msg='num candidates > 0')
self.assertGreater(int(one_row[7]), 0, msg='num examples > 0')
def test_conversion_to_tfrecord_and_back(self, original_input_file):
"""Test conversion from a native file format to tfrecord.gz, then back."""
input_path = test_utils.genomics_core_testdata(original_input_file)
tfrecord_output_path = test_utils.test_tmpfile(original_input_file +
".tfrecord.gz")
native_output_path = test_utils.test_tmpfile(original_input_file)
# Test conversion from native format to tfrecord.
self._convert(input_path, tfrecord_output_path)
# redacted
if native_output_path.endswith(".sam"):
raise unittest.SkipTest("SAM writing not yet supported")
# Test conversion from tfrecord format back to native format. Ensure that
# conversions where we would need a header, but don't have one from the
# input, trigger an error message.
if any(
native_output_path.endswith(ext)
for ext in FORMATS_REQUIRING_HEADER):
with self.assertRaisesRegexp(
converter.ConversionError,
"Input file does not have a header, which is needed to construct "
"output file"):
self._convert(tfrecord_output_path, native_output_path)
else:
self._convert(tfrecord_output_path, native_output_path)
def test_prepare_inputs(self, filename_to_write, file_string_input):
source_path = test_utils.test_tmpfile(filename_to_write)
io_utils.write_tfrecords(self.examples, source_path)
# file_string_input could be a comma-separated list. Add the prefix to all
# of them, and join it back to a string.
file_string_input = ','.join(
[test_utils.test_tmpfile(f) for f in file_string_input.split(',')])
with self.test_session() as sess:
sess.run(tf.local_variables_initializer())
sess.run(tf.global_variables_initializer())
ds = call_variants.prepare_inputs(file_string_input)
_, variants, _ = data_providers.get_infer_batches(ds,
model=self.model,
batch_size=1)
seen_variants = []
try:
while True:
seen_variants.extend(sess.run(variants))
except tf.errors.OutOfRangeError:
pass
self.assertItemsEqual(self.variants,
variant_utils.decode_variants(seen_variants))
def test_make_examples_training_end2end_with_alt_aligned_pileup(
self, alt_align, expected_shape):
region = ranges.parse_literal('chr20:10,000,000-10,010,000')
FLAGS.regions = [ranges.to_literal(region)]
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.reads = testdata.CHR20_BAM
FLAGS.candidates = test_utils.test_tmpfile(_sharded('vsc.tfrecord'))
FLAGS.examples = test_utils.test_tmpfile(_sharded('examples.tfrecord'))
FLAGS.partition_size = 1000
FLAGS.mode = 'training'
FLAGS.gvcf_gq_binsize = 5
FLAGS.alt_aligned_pileup = alt_align # This is the only input change.
FLAGS.truth_variants = testdata.TRUTH_VARIANTS_VCF
FLAGS.confident_regions = testdata.CONFIDENT_REGIONS_BED
options = make_examples.default_options(add_flags=True)
# Run make_examples with the flags above.
make_examples_core.make_examples_runner(options)
# Check the output for shape and against the golden file.
if alt_align == 'rows':
golden_file = _sharded(testdata.ALT_ALIGNED_ROWS_EXAMPLES)
elif alt_align == 'diff_channels':
golden_file = _sharded(testdata.ALT_ALIGNED_DIFF_CHANNELS_EXAMPLES)
else:
raise ValueError("Golden data doesn't exist for this alt_align option: "
'{}'.format(alt_align))
# Verify that the variants in the examples are all good.
examples = self.verify_examples(
FLAGS.examples, region, options, verify_labels=True)
self.assertDeepVariantExamplesEqual(
examples, list(tfrecord.read_tfrecords(golden_file)))
# Pileup image should have 3 rows of height 100, so resulting height is 300.
self.assertEqual(decode_example(examples[0])['image/shape'], expected_shape)
def _get_examples(use_confident_regions=False):
# `flag_name` can be either 'confident_regions' or 'regions'. Both should
# be used to constrain the set of candidates generated, and as a result
# generating the same examples.
bed_path = test_utils.test_tmpfile('vcf_candidate_importer.bed')
with gfile.Open(bed_path, 'w') as fout:
fout.write('\t'.join(['chr20', '10000000', '10001000']) + '\n')
if use_confident_regions:
FLAGS.confident_regions = bed_path
FLAGS.regions = ''
else:
FLAGS.confident_regions = ''
FLAGS.regions = bed_path
FLAGS.examples = test_utils.test_tmpfile(
_sharded('vcf_candidate_importer.tfrecord'))
FLAGS.mode = 'training'
FLAGS.reads = testdata.CHR20_BAM
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.truth_variants = testdata.TRUTH_VARIANTS_VCF
FLAGS.variant_caller = 'vcf_candidate_importer'
options = make_examples.default_options(add_flags=True)
make_examples_core.make_examples_runner(options)
# Verify that the variants in the examples are all good.
examples = self.verify_examples(
FLAGS.examples, None, options, verify_labels=False)
return examples
def test_make_examples_end2end_vcf_candidate_importer(self, mode):
FLAGS.variant_caller = 'vcf_candidate_importer'
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.reads = testdata.CHR20_BAM
FLAGS.candidates = test_utils.test_tmpfile(
_sharded('vcf_candidate_importer.{}.tfrecord'.format(mode)))
FLAGS.examples = test_utils.test_tmpfile(
_sharded('vcf_candidate_importer.examples.{}.tfrecord'.format(mode)))
FLAGS.mode = mode
if mode == 'calling':
golden_file = _sharded(
testdata.GOLDEN_VCF_CANDIDATE_IMPORTER_CALLING_EXAMPLES)
FLAGS.proposed_variants = testdata.VCF_CANDIDATE_IMPORTER_VARIANTS
# Adding the following flags to match how the testdata was created.
FLAGS.regions = 'chr20:59,777,000-60,000,000'
FLAGS.realign_reads = False
else:
golden_file = _sharded(
testdata.GOLDEN_VCF_CANDIDATE_IMPORTER_TRAINING_EXAMPLES)
FLAGS.truth_variants = testdata.TRUTH_VARIANTS_VCF
options = make_examples.default_options(add_flags=True)
make_examples_core.make_examples_runner(options)
# Verify that the variants in the examples are all good.
examples = self.verify_examples(
FLAGS.examples, None, options, verify_labels=mode == 'training')
self.assertDeepVariantExamplesEqual(
examples, list(tfrecord.read_tfrecords(golden_file)))
self.assertEqual(
decode_example(examples[0])['image/shape'],
[100, 221, dv_constants.PILEUP_NUM_CHANNELS])
def testGetShapeFromExamplesPath(self, file_name_to_write,
tfrecord_path_to_match):
example = example_pb2.Example()
valid_shape = [1, 2, 3]
example.features.feature['image/shape'].int64_list.value.extend(valid_shape)
output_file = test_utils.test_tmpfile(file_name_to_write)
io_utils.write_tfrecords([example], output_file)
tf_utils.get_shape_from_examples_path(
test_utils.test_tmpfile(tfrecord_path_to_match))
def testGlobListShardedFilePatterns(self, specs, expected_files):
# First, create all expected_files so Glob will work later.
expected_full_files = [
test_utils.test_tmpfile(f, '') for f in expected_files
]
# Create the full spec names. This one doesn't create the files.
full_specs = ','.join(
[test_utils.test_tmpfile(spec) for spec in specs.split(',')])
self.assertEqual(sorted(set(expected_full_files)),
io.glob_list_sharded_file_patterns(full_specs))
def test_call_end2end_empty_first_shard(self):
# Get only up to 10 examples.
examples = list(
io_utils.read_tfrecords(
testdata.GOLDEN_CALLING_EXAMPLES, max_records=10))
empty_first_file = test_utils.test_tmpfile('empty_1st_shard-00000-of-00002')
io_utils.write_tfrecords([], empty_first_file)
second_file = test_utils.test_tmpfile('empty_1st_shard-00001-of-00002')
io_utils.write_tfrecords(examples, second_file)
self.assertCallVariantsEmitsNRecordsForRandomGuess(
test_utils.test_tmpfile('empty_1st_shard@2'), len(examples))
def test_call_end2end_empty_first_shard(self):
# Get only up to 10 examples.
examples = list(
io_utils.read_tfrecords(
testdata.GOLDEN_CALLING_EXAMPLES, max_records=10))
empty_first_file = test_utils.test_tmpfile('empty_1st_shard-00000-of-00002')
io_utils.write_tfrecords([], empty_first_file)
second_file = test_utils.test_tmpfile('empty_1st_shard-00001-of-00002')
io_utils.write_tfrecords(examples, second_file)
self.assertCallVariantsEmitsNRecordsForRandomGuess(
test_utils.test_tmpfile('empty_1st_shard@2'), len(examples))
def test_get_shape_from_examples_path(self, file_name_to_write,
tfrecord_path_to_match):
example = example_pb2.Example()
valid_shape = [1, 2, 3]
example.features.feature['image/shape'].int64_list.value.extend(valid_shape)
output_file = test_utils.test_tmpfile(file_name_to_write)
io_utils.write_tfrecords([example], output_file)
ds = data_providers.DeepVariantDataSet(
name='test_shape',
source=test_utils.test_tmpfile(tfrecord_path_to_match),
num_examples=1)
self.assertEqual(valid_shape, ds.tensor_shape)
def test_reading_empty_input_should_raise_error(self):
empty_shard_one = test_utils.test_tmpfile(
'no_records.tfrecord-00000-of-00002')
empty_shard_two = test_utils.test_tmpfile(
'no_records.tfrecord-00001-of-00002')
io_utils.write_tfrecords([], empty_shard_one)
io_utils.write_tfrecords([], empty_shard_two)
FLAGS.infile = test_utils.test_tmpfile('no_records.tfrecord@2')
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.outfile = test_utils.test_tmpfile('no_records.vcf')
with self.assertRaisesRegexp(ValueError, 'Cannot find any records in'):
postprocess_variants.main(['postprocess_variants.py'])
def test_get_shape_from_examples_path(self, file_name_to_write,
tfrecord_path_to_match):
example = example_pb2.Example()
valid_shape = [1, 2, 3]
example.features.feature['image/shape'].int64_list.value.extend(
valid_shape)
output_file = test_utils.test_tmpfile(file_name_to_write)
io_utils.write_tfrecords([example], output_file)
ds = data_providers.DeepVariantDataSet(
name='test_shape',
source=test_utils.test_tmpfile(tfrecord_path_to_match),
num_examples=1)
self.assertEqual(valid_shape, ds.tensor_shape)
def test_get_shape_from_examples_path(self, file_name_to_write,
tfrecord_path_to_match):
example = example_pb2.Example()
valid_shape = [1, 2, 3]
example.features.feature['image/shape'].int64_list.value.extend(valid_shape)
output_file = test_utils.test_tmpfile(file_name_to_write)
tfrecord.write_tfrecords([example], output_file)
ds = data_providers.DeepVariantInput(
mode=tf.estimator.ModeKeys.PREDICT,
name='test_shape',
input_file_spec=test_utils.test_tmpfile(tfrecord_path_to_match),
num_examples=1)
self.assertEqual(valid_shape, ds.tensor_shape)
def test_reading_sharded_input_with_empty_shards_does_not_crash(self):
valid_variants = io_utils.read_tfrecords(
testdata.GOLDEN_POSTPROCESS_INPUT,
proto=deepvariant_pb2.CallVariantsOutput)
empty_shard_one = test_utils.test_tmpfile(
'reading_empty_shard.tfrecord-00000-of-00002')
none_empty_shard_two = test_utils.test_tmpfile(
'reading_empty_shard.tfrecord-00001-of-00002')
io_utils.write_tfrecords([], empty_shard_one)
io_utils.write_tfrecords(valid_variants, none_empty_shard_two)
FLAGS.infile = test_utils.test_tmpfile('reading_empty_shard.tfrecord@2')
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.outfile = test_utils.test_tmpfile('calls_reading_empty_shard.vcf')
postprocess_variants.main(['postprocess_variants.py'])
def _test_dataset_config(filename, **kwargs):
"""Creates a DeepVariantDatasetConfig(**kwargs) and writes it to filename."""
dataset_config_pbtext_filename = test_utils.test_tmpfile(filename)
dataset_config = deepvariant_pb2.DeepVariantDatasetConfig(**kwargs)
data_providers.write_dataset_config_to_pbtxt(
dataset_config, dataset_config_pbtext_filename)
return dataset_config_pbtext_filename
def test_call_end2end(self, compressed_inputs):
FLAGS.infile = make_golden_dataset(compressed_inputs)
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.outfile = test_utils.test_tmpfile('calls.vcf')
FLAGS.nonvariant_site_tfrecord_path = (
testdata.GOLDEN_POSTPROCESS_GVCF_INPUT)
FLAGS.gvcf_outfile = test_utils.test_tmpfile('gvcf_calls.vcf')
postprocess_variants.main(['postprocess_variants.py'])
self.assertEqual(
tf.gfile.FastGFile(FLAGS.outfile).readlines(),
tf.gfile.FastGFile(testdata.GOLDEN_POSTPROCESS_OUTPUT).readlines())
self.assertEqual(
tf.gfile.FastGFile(FLAGS.gvcf_outfile).readlines(),
tf.gfile.FastGFile(testdata.GOLDEN_POSTPROCESS_GVCF_OUTPUT).readlines())
def setUp(self):
super(TabixTest, self).setUp()
self.input_file = test_utils.genomics_core_testdata('test_samples.vcf.gz')
self.output_file = test_utils.test_tmpfile('test_samples.vcf.gz')
shutil.copyfile(self.input_file, self.output_file)
self.tbx_index_file = self.output_file + '.tbi'
self.csi_index_file = self.output_file + '.csi'
def test_roundtrip(self,
expected_infos,
expected_fmt,
expected_fmt1,
expected_fmt2,
reader_excluded_info=None,
reader_excluded_format=None,
writer_excluded_info=None,
writer_excluded_format=None):
expected_records = [
record.format(info=info, fmt=expected_fmt, efmts1=e1,
efmts2=e2) for record, info, e1, e2 in zip(
self.record_format_strings, expected_infos,
expected_fmt1, expected_fmt2)
]
expected = self.header + ''.join(expected_records)
with vcf.VcfReader(
test_utils.genomics_core_testdata('test_py_roundtrip.vcf'),
use_index=False,
excluded_info_fields=reader_excluded_info,
excluded_format_fields=reader_excluded_format) as reader:
records = list(reader.iterate())
output_path = test_utils.test_tmpfile('test_roundtrip_tmpfile.vcf')
with vcf.VcfWriter(
output_path,
header=reader.header,
excluded_info_fields=writer_excluded_info,
excluded_format_fields=writer_excluded_format) as writer:
for record in records:
writer.write(record)
with open(output_path) as f:
actual = f.read()
self.assertEqual(actual, expected)
def test_prepare_inputs(self, filename, expand_to_file_pattern):
source_path = test_utils.test_tmpfile(filename)
io_utils.write_tfrecords(self.examples, source_path)
if expand_to_file_pattern:
# Transform foo@3 to foo-?????-of-00003.
source_path = io_utils.NormalizeToShardedFilePattern(source_path)
with self.test_session() as sess:
sess.run(tf.local_variables_initializer())
sess.run(tf.global_variables_initializer())
ds = call_variants.prepare_inputs(source_path)
_, variants, _ = data_providers.get_infer_batches(ds,
model=self.model,
batch_size=1)
seen_variants = []
try:
while True:
seen_variants.extend(sess.run(variants))
except tf.errors.OutOfRangeError:
pass
self.assertItemsEqual(self.variants,
variant_utils.decode_variants(seen_variants))
def _test_dataset_config(filename, **kwargs):
"""Creates a DeepVariantDatasetConfig(**kwargs) and writes it to filename."""
dataset_config_pbtext_filename = test_utils.test_tmpfile(filename)
dataset_config = deepvariant_pb2.DeepVariantDatasetConfig(**kwargs)
data_providers.write_dataset_config_to_pbtxt(dataset_config,
dataset_config_pbtext_filename)
return dataset_config_pbtext_filename
def test_round_trip_vcf(self, test_datum_name):
# Round-trip variants through writing and reading:
# 1. Read variants v1 from VcfReader;
# 2. Write v1 to vcf using our VcfWriter;
# 3. Read back in using VcfReader -- v2;
# 4. compare v1 and v2.
in_file = test_utils.genomics_core_testdata(test_datum_name)
out_file = test_utils.test_tmpfile('output_' + test_datum_name)
v1_reader = vcf.VcfReader(in_file, use_index=False)
v1_records = list(v1_reader.iterate())
self.assertTrue(v1_records, 'Reader failed to find records')
header = copy.deepcopy(v1_reader.header)
writer_options = variants_pb2.VcfWriterOptions()
with vcf_writer.VcfWriter.to_file(out_file, header,
writer_options) as writer:
for record in v1_records:
writer.write(record)
v2_reader = vcf.VcfReader(out_file, use_index=False)
v2_records = list(v2_reader.iterate())
self.assertEqual(v1_records, v2_records,
'Round-tripped variants not as expected')
def testModelShapes(self):
# Builds a graph.
v0 = tf.Variable([[1, 2, 3], [4, 5, 6]], dtype=tf.float32, name='v0')
v1 = tf.Variable([[[1], [2]], [[3], [4]], [[5], [6]]],
dtype=tf.float32,
name='v1')
init_all_op = tf.initialize_all_variables()
save = tf.train.Saver({'v0': v0, 'v1': v1})
save_path = test_utils.test_tmpfile('ckpt_for_debug_string')
with tf.Session() as sess:
sess.run(init_all_op)
# Saves a checkpoint.
save.save(sess, save_path)
# Model shapes without any variable requests gives you all variables.
self.assertEqual({
'v0': (2, 3),
'v1': (3, 2, 1)
}, tf_utils.model_shapes(save_path))
# Asking for v0 gives you only v0's shape.
self.assertEqual({'v0': (2, 3)},
tf_utils.model_shapes(save_path, ['v0']))
# Asking for v1 gives you only v1's shape.
self.assertEqual({'v1': (3, 2, 1)},
tf_utils.model_shapes(save_path, ['v1']))
# Verifies model_shapes() fails for non-existent tensors.
with self.assertRaisesRegexp(KeyError, 'v3'):
tf_utils.model_shapes(save_path, ['v3'])
def test_call_variants_runs_on_gpus(self, model):
call_variants.call_variants(
examples_filename=testdata.GOLDEN_CALLING_EXAMPLES,
checkpoint_path=modeling.SKIP_MODEL_INITIALIZATION_IN_TEST,
model=model,
execution_hardware='accelerator',
output_file=test_utils.test_tmpfile('zzz.tfrecord'))
def testModelNumClasses(self):
# Builds a graph.
class_variable_name = 'class_variable_name'
v0 = tf.Variable([[1, 2, 3]],
dtype=tf.int32,
name='class_variable_name')
v1 = tf.Variable([[[1], [2]], [[3], [4]], [[5], [6]]],
dtype=tf.float32,
name='v1')
init_all_op = tf.initialize_all_variables()
save = tf.train.Saver({class_variable_name: v0, 'v1': v1})
save_path = test_utils.test_tmpfile('ckpt_for_debug_classes')
with tf.Session() as sess:
sess.run(init_all_op)
# Saves a checkpoint.
save.save(sess, save_path)
# If you pass in the correct class_variable_name, you'll find the number
# of classes.
self.assertEqual(
3, tf_utils.model_num_classes(save_path, class_variable_name))
# If the class variable name doesn't existin the checkpoint, return None.
self.assertEqual(
None, tf_utils.model_num_classes(save_path,
'non-existent-var'))
# If the checkpoint doesn't exist, return none.
self.assertEqual(
None, tf_utils.model_num_classes(None, class_variable_name))
def setUp(self):
self.out_fname = test_utils.test_tmpfile('output.vcf')
self.header = variants_pb2.VcfHeader(
contigs=[
reference_pb2.ContigInfo(name='Chr1', n_bases=50, pos_in_fasta=0),
reference_pb2.ContigInfo(name='Chr2', n_bases=25, pos_in_fasta=1),
],
sample_names=['Fido', 'Spot'],
formats=[
variants_pb2.VcfFormatInfo(
id='GT', number='1', type='String', description='Genotype'),
variants_pb2.VcfFormatInfo(
id='GQ',
number='1',
type='Float',
description='Genotype Quality')
],
)
self.options = variants_pb2.VcfWriterOptions()
self.writer = vcf_writer.VcfWriter.to_file(self.out_fname, self.header,
self.options)
self.variant = test_utils.make_variant(
chrom='Chr1',
start=10,
alleles=['A', 'C'],
)
self.variant.calls.extend([
variants_pb2.VariantCall(genotype=[0, 0], call_set_name='Fido'),
variants_pb2.VariantCall(genotype=[0, 1], call_set_name='Spot'),
])
def test_call_variants_with_empty_input(self):
source_path = test_utils.test_tmpfile('empty.tfrecord')
io_utils.write_tfrecords([], source_path)
# Make sure that prepare_inputs don't crash on empty input.
call_variants.prepare_inputs(source_path,
modeling.get_model('random_guess'),
batch_size=1)
def setUp(self):
self.out_fname = test_utils.test_tmpfile('output.vcf')
self.header = variants_pb2.VcfHeader(
contigs=[
reference_pb2.ContigInfo(name='Chr1',
n_bases=50,
pos_in_fasta=0),
reference_pb2.ContigInfo(name='Chr2',
n_bases=25,
pos_in_fasta=1),
],
sample_names=['Fido', 'Spot'],
formats=[
variants_pb2.VcfFormatInfo(id='GT',
number='1',
type='String',
description='Genotype'),
variants_pb2.VcfFormatInfo(id='GQ',
number='1',
type='Float',
description='Genotype Quality')
],
)
self.options = variants_pb2.VcfWriterOptions()
self.writer = vcf_writer.VcfWriter.to_file(self.out_fname, self.header,
self.options)
self.variant = test_utils.make_variant(
chrom='Chr1',
start=10,
alleles=['A', 'C'],
)
self.variant.calls.extend([
variants_pb2.VariantCall(genotype=[0, 0], call_set_name='Fido'),
variants_pb2.VariantCall(genotype=[0, 1], call_set_name='Spot'),
])
def test_round_trip_vcf(self, test_datum_name):
# Round-trip variants through writing and reading:
# 1. Read variants v1 from VcfReader;
# 2. Write v1 to vcf using our VcfWriter;
# 3. Read back in using VcfReader -- v2;
# 4. compare v1 and v2.
in_file = test_utils.genomics_core_testdata(test_datum_name)
out_file = test_utils.test_tmpfile('output_' + test_datum_name)
v1_reader = vcf.VcfReader(in_file)
v1_records = list(v1_reader.iterate())
self.assertTrue(v1_records, 'Reader failed to find records')
header = copy.deepcopy(v1_reader.header)
writer_options = variants_pb2.VcfWriterOptions()
with vcf_writer.VcfWriter.to_file(out_file, header,
writer_options) as writer:
for record in v1_records:
writer.write(record)
v2_reader = vcf.VcfReader(out_file)
v2_records = list(v2_reader.iterate())
self.assertEqual(v1_records, v2_records,
'Round-tripped variants not as expected')
def test_roundtrip(self,
expected_infos,
expected_fmt,
expected_fmt1,
expected_fmt2,
reader_excluded_info=None,
reader_excluded_format=None,
writer_excluded_info=None,
writer_excluded_format=None):
expected_records = [
record.format(info=info, fmt=expected_fmt, efmts1=e1,
efmts2=e2) for record, info, e1, e2 in zip(
self.record_format_strings, expected_infos,
expected_fmt1, expected_fmt2)
]
expected = self.header + ''.join(expected_records)
with vcf.VcfReader(
test_utils.genomics_core_testdata('test_py_roundtrip.vcf'),
excluded_info_fields=reader_excluded_info,
excluded_format_fields=reader_excluded_format) as reader:
records = list(reader.iterate())
output_path = test_utils.test_tmpfile('test_roundtrip_tmpfile.vcf')
with vcf.VcfWriter(
output_path,
header=reader.header,
excluded_info_fields=writer_excluded_info,
excluded_format_fields=writer_excluded_format) as writer:
for record in records:
writer.write(record)
with open(output_path) as f:
actual = f.read()
self.assertEqual(actual, expected)
def test_call_variants_runs_on_gpus(self, model):
call_variants.call_variants(
examples_filename=testdata.GOLDEN_CALLING_EXAMPLES,
checkpoint_path=modeling.SKIP_MODEL_INITIALIZATION_IN_TEST,
model=model,
execution_hardware='accelerator',
output_file=test_utils.test_tmpfile('zzz.tfrecord'))
def setUp(self):
writer_options = fastq_pb2.FastqWriterOptions()
out_fname = test_utils.test_tmpfile('output.fastq')
self.writer = fastq_writer.FastqWriter.to_file(out_fname,
writer_options)
self.expected_fastq_content = [
'@NODESC:header\n',
'GATTACA\n',
'+\n',
'BB>B@FA\n',
'@M01321:49:000000000-A6HWP:1:1101:17009:2216 1:N:0:1\n',
'CGTTAGCGCAGGGGGCATCTTCACACTGGTGACAGGTAACCGCCGTAGTAAAGGTTCCGCCTTTCACT\n',
'+\n',
'AAAAABF@BBBDGGGG?FFGFGHBFBFBFABBBHGGGFHHCEFGGGGG?FGFFHEDG3EFGGGHEGHG\n',
'@FASTQ contains multiple spaces in description\n',
'CGGCTGGTCAGGCTGACATCGCCGCCGGCCTGCAGCGAGCCGCTGC\n',
'+\n',
'FAFAF;F/9;.:/;999B/9A.DFFF;-->.AAB/FC;9-@-=;=.\n',
'@FASTQ_with_trailing_space\n',
'CGG\n',
'+\n',
'FAD\n',
]
self.record = fastq_pb2.FastqRecord(id='ID',
description='desc',
sequence='ACGTAC',
quality='ABCDEF')
def test_call_end2end(self, compressed_inputs):
FLAGS.infile = make_golden_dataset(compressed_inputs)
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.outfile = test_utils.test_tmpfile('calls.vcf')
FLAGS.nonvariant_site_tfrecord_path = (
testdata.GOLDEN_POSTPROCESS_GVCF_INPUT)
FLAGS.gvcf_outfile = test_utils.test_tmpfile('gvcf_calls.vcf')
postprocess_variants.main(['postprocess_variants.py'])
self.assertEqual(
tf.gfile.FastGFile(FLAGS.outfile).readlines(),
tf.gfile.FastGFile(testdata.GOLDEN_POSTPROCESS_OUTPUT).readlines())
self.assertEqual(
tf.gfile.FastGFile(FLAGS.gvcf_outfile).readlines(),
tf.gfile.FastGFile(testdata.GOLDEN_POSTPROCESS_GVCF_OUTPUT).readlines())
def test_realigner_diagnostics(self, enabled, emit_reads):
# Make sure that by default we aren't emitting any diagnostic outputs.
dx_dir = test_utils.test_tmpfile('dx_enabled{}_emitreads_{}'.format(
enabled, emit_reads))
region_str = 'chr20:10046178-10046188'
region = ranges.parse_literal(region_str)
assembled_region_str = 'chr20:10046096-10046267'
reads, header = _get_reads_and_header(region)
self.config = realigner.realigner_config(FLAGS)
self.config.diagnostics.enabled = enabled
self.config.diagnostics.output_root = dx_dir
self.config.diagnostics.emit_realigned_reads = emit_reads
self.reads_realigner = realigner.Realigner(self.config,
self.ref_reader, header)
_, _ = self.reads_realigner.realign_reads(reads, region)
self.reads_realigner.diagnostic_logger.close(
) # Force close all resources.
if not enabled:
# Make sure our diagnostic output isn't emitted.
self.assertFalse(tf.io.gfile.exists(dx_dir))
else:
# Our root directory exists.
self.assertTrue(tf.io.gfile.isdir(dx_dir))
# We expect a realigner_metrics.csv in our rootdir with 1 entry in it.
metrics_file = os.path.join(
dx_dir,
self.reads_realigner.diagnostic_logger.metrics_filename)
self.assertTrue(tf.io.gfile.exists(metrics_file))
with tf.io.gfile.GFile(metrics_file) as fin:
rows = list(csv.DictReader(fin))
self.assertLen(rows, 1)
self.assertEqual(set(rows[0].keys()),
{'window', 'k', 'n_haplotypes', 'time'})
self.assertEqual(rows[0]['window'], assembled_region_str)
self.assertEqual(int(rows[0]['k']), 25)
self.assertTrue(int(rows[0]['n_haplotypes']), 2)
# Check that our runtime is reasonable (greater than 0, less than 10 s).
self.assertTrue(0.0 < float(rows[0]['time']) < 10.0)
# As does the subdirectory for this region.
region_subdir = os.path.join(dx_dir, assembled_region_str)
self.assertTrue(tf.io.gfile.isdir(region_subdir))
# We always have a graph.dot
self.assertTrue(
tf.io.gfile.exists(
os.path.join(
region_subdir, self.reads_realigner.diagnostic_logger.
graph_filename)))
reads_file = os.path.join(
dx_dir, region_str, self.reads_realigner.diagnostic_logger.
realigned_reads_filename)
# if emit_reads=False then file should not exist and vice versa.
self.assertEqual(emit_reads, tf.io.gfile.exists(reads_file))
def _run():
call_variants.call_variants(
examples_filename=testdata.GOLDEN_CALLING_EXAMPLES,
checkpoint_path=modeling.SKIP_MODEL_INITIALIZATION_IN_TEST,
model=self.model,
execution_hardware=hardware_env,
max_batches=1,
batch_size=1,
output_file=test_utils.test_tmpfile('zzz.tfrecord'))
def test_call_variants_with_invalid_format(self, model, bad_format):
# Read one good record from a valid file.
example = next(io_utils.read_tfrecords(testdata.GOLDEN_CALLING_EXAMPLES))
# Overwrite the image/format field to be an invalid value
# (anything but 'raw').
example.features.feature['image/format'].bytes_list.value[0] = bad_format
source_path = test_utils.test_tmpfile('make_examples_output.tfrecord')
io_utils.write_tfrecords([example], source_path)
outfile = test_utils.test_tmpfile('call_variants_invalid_format.tfrecord')
with self.assertRaises(ValueError):
call_variants.call_variants(
examples_filename=source_path,
checkpoint_path=modeling.SKIP_MODEL_INITIALIZATION_IN_TEST,
model=model,
output_file=outfile,
batch_size=1,
max_batches=1)
def test_roundtrip_writer(self, filename):
output_path = test_utils.test_tmpfile(filename)
original_reader = sam.SamReader(test_utils.genomics_core_testdata(filename))
original_records = list(original_reader.iterate())
with sam.SamWriter(output_path, header=original_reader.header) as writer:
for record in original_records:
writer.write(record)
with sam.SamReader(output_path) as new_reader:
self.assertEqual(original_records, list(new_reader.iterate()))
def _run():
call_variants.call_variants(
examples_filename=testdata.GOLDEN_CALLING_EXAMPLES,
checkpoint_path=modeling.SKIP_MODEL_INITIALIZATION_IN_TEST,
model=self.model,
execution_hardware=hardware_env,
max_batches=1,
batch_size=1,
output_file=test_utils.test_tmpfile('zzz.tfrecord'))
def test_make_examples_with_allele_frequency(self, mode):
FLAGS.mode = 'calling'
FLAGS.ref = testdata.GRCH38_FASTA
FLAGS.reads = testdata.GRCH38_CHR20_AND_21_BAM
num_shards = 1
FLAGS.examples = test_utils.test_tmpfile(
_sharded('examples.tfrecord', num_shards))
region = ranges.parse_literal('chr20:61001-62000')
FLAGS.use_allele_frequency = True
FLAGS.regions = [ranges.to_literal(region)]
if mode == 'one vcf':
FLAGS.population_vcfs = testdata.AF_VCF_CHR20_AND_21
elif mode == 'two vcfs':
FLAGS.population_vcfs = ' '.join(
[testdata.AF_VCF_CHR20, testdata.AF_VCF_CHR21])
else:
raise ValueError('Invalid mode for parameterized test.')
options = make_examples.default_options(add_flags=True)
# Run make_examples with the flags above.
make_examples_core.make_examples_runner(options)
# Verify that the variants in the examples are all good.
examples = self.verify_examples(
FLAGS.examples, region, options, verify_labels=False)
# Pileup images should have one extra channel.
self.assertEqual([100, 221, dv_constants.PILEUP_NUM_CHANNELS + 1],
decode_example(examples[0])['image/shape'])
# Test there is something in the added channel.
# Values capture whether each loci has been seen in the observed examples.
population_matched_loci = {
'chr20:61539_A': False,
'chr20:61634_G': False,
'chr20:61644_G': False
}
for example in examples:
locus_id = vis.locus_id_from_variant(vis.variant_from_example(example))
if locus_id in population_matched_loci.keys():
channels = vis.channels_from_example(example)
self.assertGreater(
np.sum(channels[dv_constants.PILEUP_NUM_CHANNELS]),
0,
msg='There should be '
'something in the %s-th channel for variant '
'%s' % (dv_constants.PILEUP_NUM_CHANNELS + 1, locus_id))
population_matched_loci[locus_id] = True
self.assertTrue(
all(population_matched_loci.values()),
msg='Check that all '
'3 sample loci appeared in the examples.')
# Check against the golden file (same for both modes).
golden_file = _sharded(testdata.GOLDEN_ALLELE_FREQUENCY_EXAMPLES)
examples_from_golden = list(tfrecord.read_tfrecords(golden_file))
self.assertDeepVariantExamplesEqual(examples_from_golden, examples)
def test_call_end2end_with_empty_shards(self):
# Get only up to 10 examples.
examples = list(
io_utils.read_tfrecords(
testdata.GOLDEN_CALLING_EXAMPLES, max_records=10))
# Write to 15 shards, which means there will be multiple empty shards.
source_path = test_utils.test_tmpfile('sharded@{}'.format(15))
io_utils.write_tfrecords(examples, source_path)
self.assertCallVariantsEmitsNRecordsForRandomGuess(source_path,
len(examples))
def test_roundtrip_writer(self, filename):
output_path = test_utils.test_tmpfile(filename)
with fastq.FastqWriter(output_path) as writer:
for record in self.records:
writer.write(record)
with fastq.FastqReader(output_path) as reader:
v2_records = list(reader.iterate())
self.assertEqual(self.records, v2_records)
def test_catches_bad_flags(self):
# Set all of the requested flag values.
region = ranges.parse_literal('chr20:10,000,000-10,010,000')
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.reads = testdata.CHR20_BAM
FLAGS.candidates = test_utils.test_tmpfile('vsc.tfrecord')
FLAGS.examples = test_utils.test_tmpfile('examples.tfrecord')
FLAGS.regions = [ranges.to_literal(region)]
FLAGS.partition_size = 1000
FLAGS.mode = 'training'
FLAGS.truth_variants = testdata.TRUTH_VARIANTS_VCF
# This is the bad flag.
FLAGS.confident_regions = ''
with mock.patch.object(logging, 'error') as mock_logging,\
mock.patch.object(sys, 'exit') as mock_exit:
make_examples.main(['make_examples.py'])
mock_logging.assert_called_once_with(
'confident_regions is required when in training mode.')
mock_exit.assert_called_once_with(errno.ENOENT)
def make_golden_dataset(compressed_inputs=False):
if compressed_inputs:
source_path = test_utils.test_tmpfile('make_golden_dataset.tfrecord.gz')
io_utils.write_tfrecords(
io_utils.read_tfrecords(testdata.GOLDEN_TRAINING_EXAMPLES), source_path)
else:
source_path = testdata.GOLDEN_TRAINING_EXAMPLES
return data_providers.DeepVariantDataSet(
name='labeled_golden',
source=source_path,
num_examples=testdata.N_GOLDEN_TRAINING_EXAMPLES)
def test_call_variants_with_no_shape(self, model):
# Read one good record from a valid file.
example = next(io_utils.read_tfrecords(testdata.GOLDEN_CALLING_EXAMPLES))
# Remove image/shape.
del example.features.feature['image/shape']
source_path = test_utils.test_tmpfile('make_examples_out_noshape.tfrecord')
io_utils.write_tfrecords([example], source_path)
with self.assertRaisesRegexp(
ValueError, 'Invalid image/shape: we expect to find an image/shape '
'field with length 3.'):
call_variants.prepare_inputs(source_path, model, batch_size=1)
def make_golden_dataset(compressed_inputs=False):
if compressed_inputs:
source_path = test_utils.test_tmpfile(
'golden.postprocess_single_site_input.tfrecord.gz')
io_utils.write_tfrecords(
io_utils.read_tfrecords(
testdata.GOLDEN_POSTPROCESS_INPUT,
proto=deepvariant_pb2.CallVariantsOutput), source_path)
else:
source_path = testdata.GOLDEN_POSTPROCESS_INPUT
return source_path
def _parse_read_with_aux_tags(self, tag_string):
# Minimal header line to create a valid SAM file.
header_lines = '@HD\tVN:1.3\tSO:coordinate\n@SQ\tSN:chr1\tLN:248956422\n'
# A single stock read we'll add our AUX fields to.
read = 'read_name\t0\tchr1\t1\t0\t3M\t*\t0\t0\tCCC\tAAA\t' + tag_string
path = test_utils.test_tmpfile('aux_tags.bam')
with gfile.FastGFile(path, 'w') as fout:
fout.write(header_lines)
fout.write(read + '\n')
with sam.SamReader(
path, use_index=False, parse_aux_fields=True) as reader:
return list(reader.iterate())
def test_catches_bad_argv(self):
# Define valid flags to ensure raise occurs due to argv issues.
FLAGS.infile = make_golden_dataset(False)
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.outfile = test_utils.test_tmpfile('nonempty_outfile.vcf')
with mock.patch.object(logging, 'error') as mock_logging,\
mock.patch.object(sys, 'exit') as mock_exit:
postprocess_variants.main(['postprocess_variants.py', 'extra_arg'])
mock_logging.assert_called_once_with(
'Command line parsing failure: postprocess_variants does not accept '
'positional arguments but some are present on the command line: '
'"[\'postprocess_variants.py\', \'extra_arg\']".')
mock_exit.assert_called_once_with(errno.ENOENT)
def test_call_variants_non_accelerated_execution_runs(self,
execution_hardware):
# This doesn't mock out the list_devices call so it's worth keeping
# despite being very similar to the parameterized test below.
outfile = test_utils.test_tmpfile('call_variants_cpu_only.tfrecord')
call_variants.call_variants(
examples_filename=testdata.GOLDEN_CALLING_EXAMPLES,
checkpoint_path=modeling.SKIP_MODEL_INITIALIZATION_IN_TEST,
model=self.model,
execution_hardware=execution_hardware,
max_batches=1,
batch_size=1,
output_file=outfile)
def assertCallVariantsEmitsNRecordsForRandomGuess(self, filename,
num_examples):
outfile = test_utils.test_tmpfile('call_variants.tfrecord')
model = modeling.get_model('random_guess')
call_variants.call_variants(
examples_filename=filename,
checkpoint_path=modeling.SKIP_MODEL_INITIALIZATION_IN_TEST,
model=model,
output_file=outfile,
batch_size=4,
max_batches=None)
call_variants_outputs = list(
io_utils.read_tfrecords(outfile, deepvariant_pb2.CallVariantsOutput))
# Check that we have the right number of output protos.
self.assertEqual(len(call_variants_outputs), num_examples)
def test_make_read_writer_tfrecords(self):
outfile = test_utils.test_tmpfile('test.tfrecord')
writer = sam.SamWriter(outfile, header=self.header)
# Test that the writer is a context manager and that we can write a read to
# it.
with writer:
writer.write(self.read1)
writer.write(self.read2)
# Our output should have exactly one read in it.
self.assertEqual([self.read1, self.read2],
list(
io_utils.read_tfrecords(outfile,
proto=reads_pb2.Read)))
def _run_tiny_training(self, model_name, dataset):
with mock.patch(
'deepvariant.data_providers.get_dataset'
) as mock_get_dataset:
mock_get_dataset.return_value = dataset
FLAGS.train_dir = test_utils.test_tmpfile(model_name)
FLAGS.batch_size = 2
FLAGS.model_name = model_name
FLAGS.save_interval_secs = 0
FLAGS.number_of_steps = 1
FLAGS.dataset_config_pbtxt = '/path/to/mock.pbtxt'
FLAGS.start_from_checkpoint = ''
model_train.parse_and_run()
# We have a checkpoint after training.
mock_get_dataset.assert_called_once_with(FLAGS.dataset_config_pbtxt)
self.assertIsNotNone(tf.train.latest_checkpoint(FLAGS.train_dir))
def test_reading_sharded_dataset(self, compressed_inputs):
golden_dataset = make_golden_dataset(compressed_inputs)
n_shards = 3
sharded_path = test_utils.test_tmpfile('sharded@{}'.format(n_shards))
io_utils.write_tfrecords(
io_utils.read_tfrecords(golden_dataset.source), sharded_path)
config_file = _test_dataset_config(
'test_sharded.pbtxt',
name='sharded_test',
tfrecord_path=sharded_path,
num_examples=golden_dataset.num_examples)
self.assertDataSetExamplesMatchExpected(
data_providers.get_dataset(config_file).get_slim_dataset(),
golden_dataset)
def write_variant_to_tempfile(self, variant):
output_path = test_utils.test_tmpfile('test.vcf')
header = variants_pb2.VcfHeader(
contigs=[reference_pb2.ContigInfo(name='20')],
sample_names=[call.call_set_name for call in variant.calls],
formats=[
variants_pb2.VcfFormatInfo(
id='DP', number='1', type='Integer', description='Read depth'),
variants_pb2.VcfFormatInfo(
id='AD',
number='R',
type='Integer',
description='Read depth for each allele')
])
writer = vcf.VcfWriter(output_path, header=header)
with writer:
writer.write(variant)
return output_path
def test_prepare_inputs(self, filename, expand_to_file_pattern):
source_path = test_utils.test_tmpfile(filename)
io_utils.write_tfrecords(self.examples, source_path)
if expand_to_file_pattern:
# Transform foo@3 to foo-?????-of-00003.
source_path = io_utils.NormalizeToShardedFilePattern(source_path)
with self.test_session() as sess:
_, variants, _ = call_variants.prepare_inputs(
source_path, self.model, batch_size=1)
sess.run(tf.local_variables_initializer())
sess.run(tf.global_variables_initializer())
seen_variants = []
try:
while True:
seen_variants.extend(sess.run(variants))
except tf.errors.OutOfRangeError:
pass
self.assertItemsEqual(self.variants,
variant_utils.decode_variants(seen_variants))
def test_call_end2end_zero_record_file(self):
zero_record_file = test_utils.test_tmpfile('zero_record_file')
io_utils.write_tfrecords([], zero_record_file)
self.assertCallVariantsEmitsNRecordsForRandomGuess(
test_utils.test_tmpfile('zero_record_file'), 0)
def test_call_end2end(self, model, shard_inputs, include_debug_info):
FLAGS.include_debug_info = include_debug_info
examples = list(io_utils.read_tfrecords(testdata.GOLDEN_CALLING_EXAMPLES))
if shard_inputs:
# Create a sharded version of our golden examples.
source_path = test_utils.test_tmpfile('sharded@{}'.format(3))
io_utils.write_tfrecords(examples, source_path)
else:
source_path = testdata.GOLDEN_CALLING_EXAMPLES
batch_size = 4
if model.name == 'random_guess':
# For the random guess model we can run everything.
max_batches = None
else:
# For all other models we only run a single batch for inference.
max_batches = 1
outfile = test_utils.test_tmpfile('call_variants.tfrecord')
call_variants.call_variants(
examples_filename=source_path,
checkpoint_path=modeling.SKIP_MODEL_INITIALIZATION_IN_TEST,
model=model,
output_file=outfile,
batch_size=batch_size,
max_batches=max_batches)
call_variants_outputs = list(
io_utils.read_tfrecords(outfile, deepvariant_pb2.CallVariantsOutput))
# Check that we have the right number of output protos.
self.assertEqual(
len(call_variants_outputs), batch_size * max_batches
if max_batches else len(examples))
# Check that our CallVariantsOutput (CVO) have the following critical
# properties:
# - we have one CVO for each example we processed.
# - the variant in the CVO is exactly what was in the example.
# - the alt_allele_indices of the CVO match those of its corresponding
# example.
# - there are 3 genotype probabilities and these are between 0.0 and 1.0.
# We can only do this test when processing all of the variants (max_batches
# is None), since we processed all of the examples with that model.
if max_batches is None:
self.assertItemsEqual([cvo.variant for cvo in call_variants_outputs],
[tf_utils.example_variant(ex) for ex in examples])
# Check the CVO debug_info: not filled if include_debug_info is False;
# else, filled by logic based on CVO.
if not include_debug_info:
for cvo in call_variants_outputs:
self.assertEqual(cvo.debug_info,
deepvariant_pb2.CallVariantsOutput.DebugInfo())
else:
for cvo in call_variants_outputs:
self.assertEqual(cvo.debug_info.has_insertion,
variant_utils.has_insertion(cvo.variant))
self.assertEqual(cvo.debug_info.has_deletion,
variant_utils.has_deletion(cvo.variant))
self.assertEqual(cvo.debug_info.is_snp, variant_utils.is_snp(
cvo.variant))
self.assertEqual(cvo.debug_info.predicted_label,
np.argmax(cvo.genotype_probabilities))
def example_matches_call_variants_output(example, call_variants_output):
return (tf_utils.example_variant(example) == call_variants_output.variant
and tf_utils.example_alt_alleles_indices(
example) == call_variants_output.alt_allele_indices.indices)
for call_variants_output in call_variants_outputs:
# Find all matching examples.
matches = [
ex for ex in examples
if example_matches_call_variants_output(ex, call_variants_output)
]
# We should have exactly one match.
self.assertEqual(len(matches), 1)
example = matches[0]
# Check that we've faithfully copied in the alt alleles (though currently
# as implemented we find our example using this information so it cannot
# fail). Included here in case that changes in the future.
self.assertEqual(
list(tf_utils.example_alt_alleles_indices(example)),
list(call_variants_output.alt_allele_indices.indices))
# We should have exactly three genotype probabilities (assuming our
# ploidy == 2).
self.assertEqual(len(call_variants_output.genotype_probabilities), 3)
# These are probabilities so they should be between 0 and 1.
self.assertTrue(
0 <= gp <= 1 for gp in call_variants_output.genotype_probabilities)
def test_make_examples_end2end(self, mode, num_shards,
labeler_algorithm=None):
self.maxDiff = None
self.assertIn(mode, {'calling', 'training'})
region = ranges.parse_literal('chr20:10,000,000-10,010,000')
FLAGS.ref = testdata.CHR20_FASTA
FLAGS.reads = testdata.CHR20_BAM
FLAGS.candidates = test_utils.test_tmpfile(
_sharded('vsc.tfrecord', num_shards))
FLAGS.examples = test_utils.test_tmpfile(
_sharded('examples.tfrecord', num_shards))
FLAGS.regions = [ranges.to_literal(region)]
FLAGS.partition_size = 1000
FLAGS.mode = mode
FLAGS.gvcf_gq_binsize = 5
if labeler_algorithm is not None:
FLAGS.labeler_algorithm = labeler_algorithm
if mode == 'calling':
FLAGS.gvcf = test_utils.test_tmpfile(
_sharded('gvcf.tfrecord', num_shards))
else:
FLAGS.truth_variants = testdata.TRUTH_VARIANTS_VCF
FLAGS.confident_regions = testdata.CONFIDENT_REGIONS_BED
for task_id in range(max(num_shards, 1)):
FLAGS.task = task_id
options = make_examples.default_options(add_flags=True)
make_examples.make_examples_runner(options)
# Test that our candidates are reasonable, calling specific helper functions
# to check lots of properties of the output.
candidates = sorted(
io_utils.read_tfrecords(
FLAGS.candidates, proto=deepvariant_pb2.DeepVariantCall),
key=lambda c: variant_utils.variant_range_tuple(c.variant))
self.verify_deepvariant_calls(candidates, options)
self.verify_variants(
[call.variant for call in candidates], region, options, is_gvcf=False)
# Verify that the variants in the examples are all good.
examples = self.verify_examples(
FLAGS.examples, region, options, verify_labels=mode == 'training')
example_variants = [tf_utils.example_variant(ex) for ex in examples]
self.verify_variants(example_variants, region, options, is_gvcf=False)
# Verify the integrity of the examples and then check that they match our
# golden labeled examples. Note we expect the order for both training and
# calling modes to produce deterministic order because we fix the random
# seed.
if mode == 'calling':
golden_file = _sharded(testdata.GOLDEN_CALLING_EXAMPLES, num_shards)
else:
golden_file = _sharded(testdata.GOLDEN_TRAINING_EXAMPLES, num_shards)
self.assertDeepVariantExamplesEqual(
examples, list(io_utils.read_tfrecords(golden_file)))
if mode == 'calling':
nist_reader = vcf.VcfReader(testdata.TRUTH_VARIANTS_VCF)
nist_variants = list(nist_reader.query(region))
self.verify_nist_concordance(example_variants, nist_variants)
# Check the quality of our generated gvcf file.
gvcfs = variant_utils.sorted_variants(
io_utils.read_tfrecords(FLAGS.gvcf, proto=variants_pb2.Variant))
self.verify_variants(gvcfs, region, options, is_gvcf=True)
self.verify_contiguity(gvcfs, region)
gvcf_golden_file = _sharded(testdata.GOLDEN_POSTPROCESS_GVCF_INPUT,
num_shards)
expected_gvcfs = list(
io_utils.read_tfrecords(gvcf_golden_file, proto=variants_pb2.Variant))
self.assertItemsEqual(gvcfs, expected_gvcfs)
def test_call_variants_with_empty_input(self):
source_path = test_utils.test_tmpfile('empty.tfrecord')
io_utils.write_tfrecords([], source_path)
# Make sure that prepare_inputs don't crash on empty input.
call_variants.prepare_inputs(
source_path, modeling.get_model('random_guess'), batch_size=1)
def write_test_protos(self, filename): protos = [reference_pb2.ContigInfo(name=str(i)) for i in range(10)] path = test_utils.test_tmpfile(filename) io.write_tfrecords(protos, path) return protos, path
