def test_use_hp_information_conflicts(self, use_hp_information,
make_examples_extra_args,
has_conflict):
"""Confirms that PacBio use_hp_information can conflict with HP args."""
FLAGS.model_type = 'PACBIO'
FLAGS.ref = 'your_ref'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.sample_name_parent2 = 'your_sample_parent2'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.reads_parent2 = 'your_bam_parent2'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_vcf_parent2 = 'your_vcf_parent2'
FLAGS.output_gvcf_child = 'your_gvcf_child'
FLAGS.output_gvcf_parent1 = 'your_gvcf_parent1'
FLAGS.output_gvcf_parent2 = 'your_gvcf_parent2'
FLAGS.num_shards = 64
FLAGS.regions = None
FLAGS.use_hp_information = use_hp_information
FLAGS.make_examples_extra_args = make_examples_extra_args
if has_conflict:
with six.assertRaisesRegex(self, ValueError,
'conflicts with other flags'):
run_deeptrio.create_all_commands('/tmp/deeptrio_tmp_output')
else:
# Otherwise, the command should run without rasing errors.
run_deeptrio.create_all_commands('/tmp/deeptrio_tmp_output')
def test_postprocess_variants_duplicate_extra_args(
self, vcf_stats_report, postprocess_variants_extra_args,
expected_vcf_stats_report):
FLAGS.model_type = 'WGS'
FLAGS.ref = 'your_ref'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.sample_name_parent2 = 'your_sample_parent2'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.reads_parent2 = 'your_bam_parent2'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_vcf_parent2 = 'your_vcf_parent2'
FLAGS.output_gvcf_child = 'your_gvcf_child'
FLAGS.output_gvcf_parent1 = 'your_gvcf_parent1'
FLAGS.output_gvcf_parent2 = 'your_gvcf_parent2'
FLAGS.num_shards = 64
FLAGS.vcf_stats_report = vcf_stats_report
FLAGS.postprocess_variants_extra_args = postprocess_variants_extra_args
_, commands_post_process = run_deeptrio.create_all_commands(
'/tmp/deeptrio_tmp_output')
self.assertEqual(
commands_post_process[0],
'time /opt/deepvariant/bin/postprocess_variants '
'--ref "your_ref" '
'--infile '
'"/tmp/deeptrio_tmp_output/call_variants_output_child.tfrecord.gz" '
'--outfile "your_vcf_child" '
'--nonvariant_site_tfrecord_path '
'"/tmp/deeptrio_tmp_output/[email protected]" '
'--gvcf_outfile "your_gvcf_child" '
'%s' % expected_vcf_stats_report)
def test_call_variants_extra_args(self, call_variants_extra_args,
expected_args):
FLAGS.model_type = 'WGS'
FLAGS.ref = 'your_ref'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.sample_name_parent2 = 'your_sample_parent2'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.reads_parent2 = 'your_bam_parent2'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_vcf_parent2 = 'your_vcf_parent2'
FLAGS.output_gvcf_child = 'your_gvcf_child'
FLAGS.output_gvcf_parent1 = 'your_gvcf_parent1'
FLAGS.output_gvcf_parent2 = 'your_gvcf_parent2'
FLAGS.num_shards = 64
FLAGS.call_variants_extra_args = call_variants_extra_args
commands, _ = run_deeptrio.create_all_commands(
'/tmp/deeptrio_tmp_output')
self.assertEqual(
commands[1], 'time /opt/deepvariant/bin/call_variants '
'--outfile '
'"/tmp/deeptrio_tmp_output/call_variants_output_child.tfrecord.gz" '
'--examples "/tmp/deeptrio_tmp_output/[email protected]" '
'--checkpoint "/opt/models/deeptrio/wgs/child/model.ckpt" '
'%s' % expected_args)
def test_make_examples_regions(self, regions, expected_args):
FLAGS.model_type = 'WGS'
FLAGS.ref = 'your_ref'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.sample_name_parent2 = 'your_sample_parent2'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.reads_parent2 = 'your_bam_parent2'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_vcf_parent2 = 'your_vcf_parent2'
FLAGS.num_shards = 64
FLAGS.regions = regions
commands, _ = run_deeptrio.create_all_commands(
'/tmp/deeptrio_tmp_output')
self.assertEqual(
commands[0], 'time seq 0 63 | parallel -q --halt 2 --line-buffer '
'/opt/deepvariant/bin/deeptrio/make_examples --mode calling '
'--ref "your_ref" --reads_parent1 "your_bam_parent1" '
'--reads_parent2 "your_bam_parent2" '
'--reads "your_bam_child" '
'--examples "/tmp/deeptrio_tmp_output/[email protected]" '
'--sample_name "your_sample_child" '
'--sample_name_parent1 "your_sample_parent1" '
'--sample_name_parent2 "your_sample_parent2" '
'%s '
'--task {}' % expected_args)
def test_pacbio_args_overwrite(self, make_examples_extra_args,
expected_args):
"""Confirms that adding extra flags can overwrite the default from mode."""
FLAGS.model_type = 'PACBIO'
FLAGS.ref = 'your_ref'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.sample_name_parent2 = 'your_sample_parent2'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.reads_parent2 = 'your_bam_parent2'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_vcf_parent2 = 'your_vcf_parent2'
FLAGS.output_gvcf_child = 'your_gvcf_child'
FLAGS.output_gvcf_parent1 = 'your_gvcf_parent1'
FLAGS.output_gvcf_parent2 = 'your_gvcf_parent2'
FLAGS.num_shards = 64
FLAGS.regions = None
FLAGS.make_examples_extra_args = make_examples_extra_args
commands, _ = run_deeptrio.create_all_commands(
'/tmp/deeptrio_tmp_output')
self.assertEqual(
commands[0], 'time seq 0 63 | parallel -q --halt 2 --line-buffer '
'/opt/deepvariant/bin/deeptrio/make_examples --mode calling '
'--ref "your_ref" --reads_parent1 "your_bam_parent1" '
'--reads_parent2 "your_bam_parent2" '
'--reads "your_bam_child" '
'--examples "/tmp/deeptrio_tmp_output/[email protected]" '
'--sample_name "your_sample_child" '
'--sample_name_parent1 "your_sample_parent1" '
'--sample_name_parent2 "your_sample_parent2" '
'%s'
'--task {}' % expected_args)
def test_duo_make_examples_commands_with_types(self, model_type,
extra_args_plus_gvcf):
FLAGS.model_type = model_type
FLAGS.ref = 'your_ref'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_gvcf_child = 'your_gvcf_child'
FLAGS.output_gvcf_parent1 = 'your_gvcf_parent1'
FLAGS.output_gvcf_merged = 'your_gvcf_merged'
FLAGS.num_shards = 64
commands, _ = run_deeptrio.create_all_commands(
'/tmp/deeptrio_tmp_output')
self.assertEqual(
commands[0], 'time seq 0 63 '
'| parallel -q --halt 2 --line-buffer '
'/opt/deepvariant/bin/deeptrio/make_examples '
'--mode calling '
'--ref "your_ref" '
'--reads_parent1 "your_bam_parent1" '
'--reads "your_bam_child" '
'--examples "/tmp/deeptrio_tmp_output/[email protected]" '
'--sample_name "your_sample_child" '
'--sample_name_parent1 "your_sample_parent1" '
'%s'
'--task {}' % extra_args_plus_gvcf)
def _create_all_commands_and_check_stdout(self, expected_stdout=None):
with mock.patch('sys.stdout', new_callable=io.StringIO) as mock_stdout:
commands, postprocess_cmds = run_deeptrio.create_all_commands(
'/tmp/deeptrio_tmp_output')
# Confirm that these basic commands don't have extra messages printed out
# to stdout.
if expected_stdout is None:
self.assertEmpty(mock_stdout.getvalue())
else:
self.assertEqual(mock_stdout.getvalue(), expected_stdout)
return commands, postprocess_cmds
def test_duo_call_variants_postprocess_variants_commands(self, model_type):
FLAGS.model_type = model_type
FLAGS.ref = 'your_ref'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_gvcf_child = 'your_gvcf_child'
FLAGS.output_gvcf_parent1 = 'your_gvcf_parent1'
FLAGS.output_gvcf_merged = 'your_gvcf_merged'
FLAGS.num_shards = 64
commands, post_process_commands = run_deeptrio.create_all_commands(
'/tmp/deeptrio_tmp_output')
self.assertEqual(
commands[1], 'time /opt/deepvariant/bin/call_variants '
'--outfile '
'"/tmp/deeptrio_tmp_output/call_variants_output_child.tfrecord.gz" '
'--examples "/tmp/deeptrio_tmp_output/[email protected]" '
'--checkpoint "/opt/models/deeptrio/{}/child/model.ckpt"'.format(
model_type.lower()))
self.assertEqual(
commands[2], 'time /opt/deepvariant/bin/call_variants '
'--outfile '
'"/tmp/deeptrio_tmp_output/call_variants_output_parent1.tfrecord.gz" '
'--examples "/tmp/deeptrio_tmp_output/[email protected]" '
'--checkpoint "/opt/models/deeptrio/{}/parent/model.ckpt"'.format(
model_type.lower()))
self.assertEqual(
post_process_commands[0],
'time /opt/deepvariant/bin/postprocess_variants '
'--ref "your_ref" '
'--infile '
'"/tmp/deeptrio_tmp_output/call_variants_output_child.tfrecord.gz" '
'--outfile "your_vcf_child" '
'--nonvariant_site_tfrecord_path '
'"/tmp/deeptrio_tmp_output/[email protected]" '
'--gvcf_outfile "your_gvcf_child"')
self.assertEqual(
post_process_commands[1],
'time /opt/deepvariant/bin/postprocess_variants '
'--ref "your_ref" '
'--infile '
'"/tmp/deeptrio_tmp_output/call_variants_output_parent1.tfrecord.gz" '
'--outfile "your_vcf_parent1" '
'--nonvariant_site_tfrecord_path '
'"/tmp/deeptrio_tmp_output/[email protected]" '
'--gvcf_outfile "your_gvcf_parent1"')
# pylint: disable=g-generic-assert
self.assertLen(commands, 3)
self.assertLen(post_process_commands, 2)
def test_use_hp_information_only_with_pacbio(self, model_type):
"""Confirms use_hp_information only works for."""
FLAGS.model_type = model_type
FLAGS.ref = 'your_ref'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.sample_name_parent2 = 'your_sample_parent2'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.reads_parent2 = 'your_bam_parent2'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_vcf_parent2 = 'your_vcf_parent2'
FLAGS.output_gvcf_child = 'your_gvcf_child'
FLAGS.output_gvcf_parent1 = 'your_gvcf_parent1'
FLAGS.output_gvcf_parent2 = 'your_gvcf_parent2'
FLAGS.num_shards = 64
FLAGS.regions = None
FLAGS.use_hp_information = True
with six.assertRaisesRegex(
self, ValueError, '--use_hp_information can only be used with '
'--model_type="PACBIO"'):
run_deeptrio.create_all_commands('/tmp/deeptrio_tmp_output')
def test_make_examples_extra_args_invalid(self):
FLAGS.model_type = 'WGS'
FLAGS.ref = 'your_ref'
FLAGS.sample_name_child = 'your_sample_child'
FLAGS.sample_name_parent1 = 'your_sample_parent1'
FLAGS.sample_name_parent2 = 'your_sample_parent2'
FLAGS.reads_child = 'your_bam_child'
FLAGS.reads_parent1 = 'your_bam_parent1'
FLAGS.reads_parent2 = 'your_bam_parent2'
FLAGS.output_vcf_child = 'your_vcf_child'
FLAGS.output_vcf_parent1 = 'your_vcf_parent1'
FLAGS.output_vcf_parent2 = 'your_vcf_parent2'
FLAGS.output_gvcf_child = 'your_gvcf_child'
FLAGS.output_gvcf_parent1 = 'your_gvcf_parent1'
FLAGS.output_gvcf_parent2 = 'your_gvcf_parent2'
FLAGS.num_shards = 64
FLAGS.make_examples_extra_args = 'keep_secondary_alignments'
with six.assertRaisesRegex(self, ValueError, 'not enough values to unpack'):
_, _ = run_deeptrio.create_all_commands('/tmp/deeptrio_tmp_output')
