def test_error_on_bad_config(self):
with self.assertRaises(SystemExit) as err:
main([
SUBCOMMAND.SETUP, 'thing/that/doesnot/exist.cfg', '-o',
self.temp_output
])
self.assertEqual(2, err.exception.code)
def test_marker_not_enough_args(self):
with patch.object(sys, 'argv', [
'mavis', SUBCOMMAND.OVERLAY, 'GAGE4', '--annotations',
ANNOTATIONS, '--output', self.temp_output, '--marker', 'm'
]):
try:
returncode = main()
except SystemExit as err:
self.assertNotEqual(0, err.code)
else:
self.assertNotEqual(0, returncode)
self.assertTrue(
glob_exists(os.path.join(self.temp_output, '*GAGE4*.svg')))
with patch.object(sys, 'argv', [
'mavis', SUBCOMMAND.OVERLAY, '--annotations', ANNOTATIONS,
'GAGE4', '--output', self.temp_output, '--marker'
]):
try:
returncode = main()
except SystemExit as err:
self.assertNotEqual(0, err.code)
else:
self.assertNotEqual(0, returncode)
self.assertTrue(
glob_exists(os.path.join(self.temp_output, '*GAGE4*.svg')))
def test_clean_files(self):
args = [
'mavis', SUBCOMMAND.PIPELINE, CLEAN_CONFIG, '-o', self.temp_output
]
with patch.object(sys, 'argv', args):
self.assertEqual(0, main())
# check that the subdirectories were built
for lib in [MOCK_GENOME + '_*', MOCK_TRANS + '_*']:
self.check_cluster(lib)
# run validation
self.assertTrue(
glob_exists(self.temp_output, lib, SUBCOMMAND.VALIDATE))
qsub = unique_exists(
os.path.join(self.temp_output, lib, SUBCOMMAND.VALIDATE,
'*-1/submit.sh'))
args = convert_qsub_to_args(
qsub) # read the arguments from the file
print(args)
with patch.object(sys, 'argv', args):
self.assertEqual(0, main())
for suffix in [
'evidence.bed', 'validation-failed.tab',
'validation-passed.tab'
]:
self.assertTrue(
glob_exists(self.temp_output, lib,
SUBCOMMAND.VALIDATE + '/*-1', '*.' + suffix))
for suffix in [
'contigs.bam',
'contigs.blat_out.pslx',
'contigs.fa',
'contigs.sorted.bam',
'contigs.sorted.bam.bai',
'igv.batch',
'raw_evidence.bam',
'raw_evidence.sorted.bam',
'raw_evidence.sorted.bam.bai',
]:
self.assertFalse(
glob_exists(self.temp_output, lib,
SUBCOMMAND.VALIDATE + '/*-1', '*.' + suffix))
self.assertTrue(
glob_exists(self.temp_output, lib,
SUBCOMMAND.VALIDATE + '/*-1', '*.COMPLETE'))
self.check_and_run_annotate(lib)
# now run the pairing
self.check_and_run_pairing()
# now run the summary
self.check_and_run_summary()
with patch.object(
sys, 'argv',
['mavis', SUBCOMMAND.CHECKER, '-o', self.temp_output]):
self.assertEqual(0, main())
self.assertTrue(glob_exists(self.temp_output, 'submit_pipeline*.sh'))
def test_reference_from_env(self):
os.environ.update({
'MAVIS_TEMPLATE_METADATA':
get_data('cytoBand.txt'),
'MAVIS_ANNOTATIONS':
get_data('mock_annotations.json'),
'MAVIS_MASKING':
get_data('mock_masking.tab'),
'MAVIS_REFERENCE_GENOME':
get_data('mock_reference_genome.fa'),
'MAVIS_ALIGNER_REFERENCE':
get_data('mock_reference_genome.2bit'),
'MAVIS_DGV_ANNOTATION':
get_data('mock_dgv_annotation.txt'),
})
main([
SUBCOMMAND.SETUP,
get_data('reference_from_env.cfg'), '-o', self.temp_output
])
self.assertTrue(glob_exists(self.temp_output, 'build.cfg'))
main([SUBCOMMAND.SCHEDULE, '-o', self.temp_output, '--submit'])
# check that the subdirectories were built
for lib in [MOCK_GENOME + '_*']:
self.check_cluster(lib)
self.check_validate(lib)
self.check_aligner_output_files(lib)
self.check_annotate(lib)
self.check_pairing()
self.check_summary(count=2)
def test_slurm_skip_cluster(self):
os.environ['MAVIS_SCHEDULER'] = 'SLURM'
args = [
SUBCOMMAND.SETUP,
CONFIG,
'-o',
self.temp_output,
'--skip_stage',
SUBCOMMAND.CLUSTER,
]
main(args)
self.assertTrue(glob_exists(self.temp_output, 'build.cfg'))
self.assertTrue(
glob_exists(self.temp_output,
'*',
SUBCOMMAND.VALIDATE,
'submit.sh',
n=2))
self.assertTrue(
glob_exists(self.temp_output,
'*',
SUBCOMMAND.ANNOTATE,
'submit.sh',
n=2))
self.assertTrue(
glob_exists(self.temp_output, SUBCOMMAND.PAIR, 'submit.sh'))
self.assertTrue(
glob_exists(self.temp_output, SUBCOMMAND.SUMMARY, 'submit.sh'))
def test_no_optional_files(self):
args = [
'mavis', SUBCOMMAND.PIPELINE,
data('no_opt_pipeline.cfg'), '-o', self.temp_output
]
with patch.object(sys, 'argv', args):
self.assertEqual(0, main())
# check that the subdirectories were built
for lib in [MOCK_GENOME + '_*', MOCK_TRANS + '_*']:
self.check_cluster(lib)
self.check_and_run_validate(lib)
self.check_aligner_output_files(lib)
self.check_and_run_annotate(lib)
# now run the pairing
self.check_and_run_pairing()
# now run the summary
self.check_and_run_summary()
with patch.object(
sys, 'argv',
['mavis', SUBCOMMAND.CHECKER, '-o', self.temp_output]):
self.assertEqual(0, main())
self.assertTrue(glob_exists(self.temp_output, 'submit_pipeline*.sh'))
def test_missing_reference(self):
args = [
'mavis', SUBCOMMAND.PIPELINE,
data('missing_reference.cfg'), '-o', self.temp_output
]
with patch.object(sys, 'argv', args):
with self.assertRaises(OSError):
main()
def test_error_on_bad_input_file(self):
args = [
'mavis', SUBCOMMAND.PIPELINE,
data('bad_input_file.cfg'), '-o', self.temp_output
]
with patch.object(sys, 'argv', args):
with self.assertRaises(FileNotFoundError):
main()
def test_error_on_bad_config(self):
args = [
'mavis', SUBCOMMAND.PIPELINE, 'thing/that/doesnot/exist.cfg', '-o',
self.temp_output
]
with patch.object(sys, 'argv', args):
print(sys.argv)
try:
main()
raise AssertionError('should have thrown error')
except SystemExit as err:
self.assertEqual(2, err.code)
def test_pipeline_with_bwa(self):
main([SUBCOMMAND.SETUP, BWA_CONFIG, '-o', self.temp_output])
self.assertTrue(glob_exists(self.temp_output, 'build.cfg'))
main([SUBCOMMAND.SCHEDULE, '-o', self.temp_output, '--submit'])
# check that the subdirectories were built
for lib in [MOCK_GENOME + '_*', MOCK_TRANS + '_*']:
self.check_cluster(lib)
self.check_validate(lib)
self.check_aligner_output_files(lib, mem=True)
self.check_annotate(lib)
self.check_pairing()
self.check_summary()
def test_skip_clustering(self):
main([
SUBCOMMAND.SETUP, CONFIG, '-o', self.temp_output, '--skip_stage',
SUBCOMMAND.CLUSTER
])
self.assertTrue(glob_exists(self.temp_output, 'build.cfg'))
main([SUBCOMMAND.SCHEDULE, '-o', self.temp_output, '--submit'])
# check that the subdirectories were built
for lib in [MOCK_GENOME + '_*', MOCK_TRANS + '_*']:
self.check_cluster(lib, skipped=True)
self.check_validate(lib)
self.check_aligner_output_files(lib)
self.check_annotate(lib)
self.check_pairing()
self.check_summary()
def check_and_run_validate(self, lib):
# run validation
self.assertTrue(glob_exists(self.temp_output, lib,
SUBCOMMAND.VALIDATE))
qsub = unique_exists(
os.path.join(self.temp_output, lib, SUBCOMMAND.VALIDATE,
'*-1/submit.sh'))
args = convert_qsub_to_args(qsub) # read the arguments from the file
print(args)
with patch.object(sys, 'argv', args):
self.assertEqual(0, main())
for suffix in [
'contigs.bam', 'contigs.fa', 'contigs.sorted.bam',
'contigs.sorted.bam.bai', 'evidence.bed', 'igv.batch',
'raw_evidence.bam', 'raw_evidence.sorted.bam',
'raw_evidence.sorted.bam.bai', 'validation-failed.tab',
'validation-passed.tab'
]:
self.assertTrue(
glob_exists(self.temp_output, lib,
SUBCOMMAND.VALIDATE + '/*-1', '*.' + suffix))
self.assertTrue(
glob_exists(self.temp_output, lib, SUBCOMMAND.VALIDATE + '/*-1',
'*.COMPLETE'))
def test_read_depth_plot_not_binned_but_stranded(self, config_json,
output_dir):
# no ymax
with patch.object(
sys,
'argv',
[
'mavis',
SUBCOMMAND.OVERLAY,
'GAGE4',
'--config',
config_json,
'--output',
output_dir,
'--read_depth_plot',
'axis',
BAM,
'1',
'none',
'True',
],
):
try:
returncode = main()
except SystemExit as err:
assert err.code == 0
else:
assert returncode is None
assert glob_exists(os.path.join(output_dir, '*GAGE4*.svg'))
def test_marker_range(self, config_json, output_dir):
with patch.object(
sys,
'argv',
[
'mavis',
SUBCOMMAND.OVERLAY,
'GAGE4',
'--config',
config_json,
'--output',
output_dir,
'--marker',
'm',
'49364900',
'49365900',
],
):
try:
returncode = main()
except SystemExit as err:
assert err.code == 0
else:
assert returncode is None
assert glob_exists(os.path.join(output_dir, '*GAGE4*.svg'))
def test_skip_validation(self):
main([
SUBCOMMAND.SETUP, CONFIG, '-o', self.temp_output, '--skip_stage',
SUBCOMMAND.VALIDATE
])
self.assertTrue(glob_exists(self.temp_output, 'build.cfg'))
main([SUBCOMMAND.SCHEDULE, '-o', self.temp_output, '--submit'])
# check that the subdirectories were built
for lib in [MOCK_GENOME + '_*', MOCK_TRANS + '_*']:
self.check_cluster(lib)
self.assertFalse(
glob_exists(self.temp_output, lib, SUBCOMMAND.VALIDATE))
self.check_annotate(lib)
self.check_pairing()
self.check_summary()
def test_convert(self):
with patch.object(sys, 'argv', ['mavis', SUBCOMMAND.CONVERT, '-h']):
try:
returncode = main()
except SystemExit as err:
assert err.code == 0
else:
assert returncode == 0
def test_main(self):
with patch.object(sys, 'argv', ['mavis', '-h']):
try:
returncode = main()
except SystemExit as err:
assert err.code == 0
else:
assert returncode == 0
def test_summary(self):
with patch.object(sys, 'argv', ['mavis', SUBCOMMAND.SUMMARY, '-h']):
try:
returncode = main()
except SystemExit as err:
assert err.code == 0
else:
assert returncode == 0
def test_bad_option(self):
with patch.object(sys, 'argv', ['mavis', SUBCOMMAND.SETUP, '--blargh']):
try:
returncode = main()
except SystemExit as err:
self.assertNotEqual(0, err.code)
else:
self.assertNotEqual(0, returncode)
def test_pipeline(self):
with patch.object(sys, 'argv', ['mavis', SUBCOMMAND.PIPELINE, '-h']):
try:
returncode = main()
except SystemExit as err:
self.assertEqual(0, err.code)
else:
self.assertEqual(0, returncode)
def test_overlay(self):
with patch.object(sys, 'argv', ['mavis', SUBCOMMAND.OVERLAY, '-h']):
try:
returncode = main()
except SystemExit as err:
self.assertEqual(0, err.code)
else:
self.assertEqual(0, returncode)
def test_annotate(self):
with patch.object(sys, 'argv', ['mavis', SUBCOMMAND.ANNOTATE, '-h']):
try:
returncode = main()
except SystemExit as err:
self.assertEqual(0, err.code)
else:
self.assertEqual(0, returncode)
def test_validate(self):
with patch.object(sys, 'argv', ['mavis', SUBCOMMAND.VALIDATE, '-h']):
try:
returncode = main()
except SystemExit as err:
assert err.code == 0
else:
assert returncode == 0
def test_ref_alt_count(self):
with patch.object(sys, 'argv', ['calculate_ref_alt_counts', '-h']):
try:
returncode = main()
except SystemExit as err:
self.assertEqual(0, err.code)
else:
self.assertEqual(0, returncode)
def test_cluster(self):
with patch.object(sys, 'argv', ['mavis', SUBCOMMAND.CLUSTER, '-h']):
try:
returncode = main()
except SystemExit as err:
self.assertEqual(0, err.code)
else:
self.assertEqual(0, returncode)
def test_main(self):
with patch.object(sys, 'argv', ['mavis', '-h']):
try:
returncode = main()
except SystemExit as err:
self.assertEqual(0, err.code)
else:
self.assertEqual(0, returncode)
def test_bad_option(self):
with patch.object(sys, 'argv',
['mavis', SUBCOMMAND.SETUP, '--blargh']):
try:
returncode = main()
except SystemExit as err:
assert err.code != 0
else:
assert returncode != 0
def run_main(self, exit_status=0):
outputfile = os.path.join(self.temp_output, 'config.cfg')
self.args.extend(['-w', outputfile])
with mock.patch.object(sys, 'argv', [str(a) for a in self.args]):
print('sys.argv', sys.argv)
try:
return_code = main()
except SystemExit as ex:
return_code = ex.code
self.assertEqual(exit_status, return_code)
def test_clean_files(self):
main([SUBCOMMAND.SETUP, CLEAN_CONFIG, '-o', self.temp_output])
self.assertTrue(glob_exists(self.temp_output, 'build.cfg'))
main([SUBCOMMAND.SCHEDULE, '-o', self.temp_output, '--submit'])
# check that the subdirectories were built
for lib in [MOCK_GENOME + '_*', MOCK_TRANS + '_*']:
self.check_cluster(lib)
self.assertTrue(
glob_exists(self.temp_output, lib, SUBCOMMAND.VALIDATE))
for suffix in [
'evidence.bed', 'validation-failed.tab',
'validation-passed.tab'
]:
self.assertTrue(
glob_exists(self.temp_output, lib,
SUBCOMMAND.VALIDATE + '/*-1', suffix))
for suffix in [
'contigs.bam',
'contigs.blat_out.pslx',
'contigs.fa',
'contigs.sorted.bam',
'contigs.sorted.bam.bai',
'igv.batch',
'raw_evidence.bam',
'raw_evidence.sorted.bam',
'raw_evidence.sorted.bam.bai',
]:
self.assertFalse(
glob_exists(self.temp_output, lib,
SUBCOMMAND.VALIDATE + '/*-1', suffix),
msg=suffix,
)
self.assertTrue(
glob_exists(self.temp_output, lib,
SUBCOMMAND.VALIDATE + '/*-1', 'MAVIS-*.COMPLETE'))
self.check_annotate(lib)
self.check_pairing()
self.check_summary()
def test_reference_from_env(self):
args = [
'mavis', SUBCOMMAND.PIPELINE,
data('reference_from_env.cfg'), '-o', self.temp_output
]
env = {k: v for k, v in os.environ.items()}
env.update({
'MAVIS_TEMPLATE_METADATA':
'tests/integration/data/cytoBand.txt',
'MAVIS_ANNOTATIONS':
'tests/integration/data/mock_annotations.json',
'MAVIS_MASKING':
'tests/integration/data/mock_masking.tab',
'MAVIS_REFERENCE_GENOME':
'tests/integration/data/mock_reference_genome.fa',
'MAVIS_ALIGNER_REFERENCE':
'tests/integration/data/mock_reference_genome.2bit',
'MAVIS_DGV_ANNOTATION':
'tests/integration/data/mock_dgv_annotation.txt',
})
with patch.object(os, 'environ', env):
with patch.object(sys, 'argv', args):
self.assertEqual(0, main())
# check that the subdirectories were built
for lib in [MOCK_GENOME + '_*']:
self.check_cluster(lib)
self.check_and_run_validate(lib)
self.check_aligner_output_files(lib)
self.check_and_run_annotate(lib)
# now run the pairing
self.check_and_run_pairing()
# now run the summary
self.check_and_run_summary(count=2)
with patch.object(
sys, 'argv',
['mavis', SUBCOMMAND.CHECKER, '-o', self.temp_output]):
self.assertEqual(0, main())
self.assertTrue(
glob_exists(self.temp_output, 'submit_pipeline*.sh'))