def test_validate_SFF(self):
prep_info = {"1.SKB2.640194": {"run_prefix": "GAX40"},
"1.SKM4.640180": {"run_prefix": "GAX40"},
"1.SKB3.640195": {"run_prefix": "GAX50"}}
files = {'raw_sff': ['/path/to/GAX401.sff', '/path/to/GAX402.sff',
'/path/to/GAX501.sff']}
job_id = self._create_template_and_job(prep_info, files, "SFF")
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, 'SFF')
self.assertTrue(obs_success)
filepaths = [('/path/to/GAX401.sff', 'raw_sff'),
('/path/to/GAX402.sff', 'raw_sff'),
('/path/to/GAX501.sff', 'raw_sff')]
exp = [ArtifactInfo(None, "SFF", filepaths)]
self.assertEqual(obs_ainfo, exp)
self.assertEqual(obs_error, "")
# let's test a failure
files = {'raw_sff': ['/path/to/GAX401.sff', '/path/to/GAX402.sff']}
job_id = self._create_template_and_job(prep_info, files, "SFF")
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, 'SFF')
error = ("Error creating artifact. Offending files:\nraw_sff: The "
"following run prefixes in the prep information file do not "
"match any file: GAX50")
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertItemsEqual(obs_error, error)
def test_validate_multiple_single_lane(self):
prep_info = {
"1.SKB2.640194": {
"not_a_run_prefix": "prefix1"
},
"1.SKM4.640180": {
"not_a_run_prefix": "prefix1"
},
"1.SKB3.640195": {
"not_a_run_prefix": "prefix2"
}
}
files = {
'raw_forward_seqs': ['/path/to/prefix1.fastq'],
'raw_barcodes': ['/path/to/prefix1_b.fastq']
}
atype = "FASTQ"
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
self.assertTrue(obs_success)
filepaths = [('/path/to/prefix1_b.fastq', 'raw_barcodes'),
('/path/to/prefix1.fastq', 'raw_forward_seqs')]
exp = [ArtifactInfo(None, atype, filepaths)]
self.assertEqual(obs_ainfo, exp)
self.assertEqual(obs_error, "")
def test_validate_multiple(self):
prep_info = {
'1.SKB2.640194': {
'run_prefix': 'prefix1'
},
'1.SKM4.640180': {
'run_prefix': 'prefix1'
},
'1.SKB3.640195': {
'run_prefix': 'prefix2'
}
}
files = {
'raw_forward_seqs':
['/path/to/prefix1.fastq', '/path/to/prefix2.fastq'],
'raw_barcodes':
['/path/to/prefix1_b.fastq', '/path/to/prefix2_b.fastq']
}
atype = "FASTQ"
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
self.assertTrue(obs_success)
filepaths = [('/path/to/prefix1_b.fastq', 'raw_barcodes'),
('/path/to/prefix2_b.fastq', 'raw_barcodes'),
('/path/to/prefix1.fastq', 'raw_forward_seqs'),
('/path/to/prefix2.fastq', 'raw_forward_seqs')]
exp = [ArtifactInfo(None, "FASTQ", filepaths)]
self.assertEqual(obs_ainfo, exp)
self.assertEqual(obs_error, "")
def test_validate_SFF(self):
prep_info = {
"1.SKB2.640194": {
"run_prefix": "GAX40"
},
"1.SKM4.640180": {
"run_prefix": "GAX40"
},
"1.SKB3.640195": {
"run_prefix": "GAX50"
}
}
files = {
'raw_sff': [
'/path/to/GAX401.sff', '/path/to/GAX402.sff',
'/path/to/GAX501.sff'
]
}
job_id = self._create_template_and_job(prep_info, files, "SFF")
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, 'SFF')
self.assertTrue(obs_success)
filepaths = [('/path/to/GAX401.sff', 'raw_sff'),
('/path/to/GAX402.sff', 'raw_sff'),
('/path/to/GAX501.sff', 'raw_sff')]
exp = [ArtifactInfo(None, "SFF", filepaths)]
self.assertEqual(obs_ainfo, exp)
self.assertEqual(obs_error, "")
# let's test a failure
files = {'raw_sff': ['/path/to/GAX401.sff', '/path/to/GAX402.sff']}
job_id = self._create_template_and_job(prep_info, files, "SFF")
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, 'SFF')
error = ("Error creating artifact. Offending files:\nraw_sff: The "
"following run prefixes in the prep information file do not "
"match any file: GAX50")
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertItemsEqual(obs_error, error)
def test_validate_multiple_single_lane(self):
prep_info = {"1.SKB2.640194": {"not_a_run_prefix": "prefix1"},
"1.SKM4.640180": {"not_a_run_prefix": "prefix1"},
"1.SKB3.640195": {"not_a_run_prefix": "prefix2"}}
files = {'raw_forward_seqs': ['/path/to/prefix1.fastq'],
'raw_barcodes': ['/path/to/prefix1_b.fastq']}
atype = "FASTQ"
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
self.assertTrue(obs_success)
filepaths = [
('/path/to/prefix1_b.fastq', 'raw_barcodes'),
('/path/to/prefix1.fastq', 'raw_forward_seqs')]
exp = [ArtifactInfo(None, atype, filepaths)]
self.assertEqual(obs_ainfo, exp)
self.assertEqual(obs_error, "")
def test_validate_multiple(self):
prep_info = {
'1.SKB2.640194': {'run_prefix': 'prefix1'},
'1.SKM4.640180': {'run_prefix': 'prefix1'},
'1.SKB3.640195': {'run_prefix': 'prefix2'}}
files = {'raw_forward_seqs': ['/path/to/prefix1.fastq',
'/path/to/prefix2.fastq'],
'raw_barcodes': ['/path/to/prefix1_b.fastq',
'/path/to/prefix2_b.fastq']}
atype = "FASTQ"
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
self.assertTrue(obs_success)
filepaths = [
('/path/to/prefix1_b.fastq', 'raw_barcodes'),
('/path/to/prefix2_b.fastq', 'raw_barcodes'),
('/path/to/prefix1.fastq', 'raw_forward_seqs'),
('/path/to/prefix2.fastq', 'raw_forward_seqs')]
exp = [ArtifactInfo(None, "FASTQ", filepaths)]
self.assertEqual(obs_ainfo, exp)
self.assertEqual(obs_error, "")
def test_validate_multiple_error(self):
# Filepath type not supported
prep_info = {"1.SKB2.640194": {"run_prefix": "prefix1"},
"1.SKM4.640180": {"run_prefix": "prefix1"},
"1.SKB3.640195": {"run_prefix": "prefix2"}}
files = {'Unknown': ['/path/to/file1.fastq']}
atype = "FASTQ"
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertEqual(obs_error,
"Filepath type(s) Unknown not supported by artifact "
"type FASTQ. Supported filepath types: raw_barcodes, "
"raw_forward_seqs, raw_reverse_seqs")
# Number of provided files != Num run prefix values
files = {'raw_forward_seqs': ['/path/to/file1.fastq'],
'raw_barcodes': ['/path/to/file1_b.fastq',
'/path/to/file2_b.fastq',
'/path/to/file3_b.fastq']}
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
error = ("Error creating artifact. Offending files:\nraw_forward_seqs:"
" The number of provided files (1) doesn't match the number "
"of run prefix values in the prep info (2): file1.fastq\n"
"raw_barcodes: The number of provided files (3) doesn't "
"match the number of run prefix values in the prep info (2): "
"file1_b.fastq, file2_b.fastq, file3_b.fastq")
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertItemsEqual(obs_error.split('\n'), error.split('\n'))
# File doesn't match any run prefix
files = {'raw_forward_seqs': ['/path/to/file1.fastq',
'/path/to/prefix2.fastq'],
'raw_barcodes': ['/path/to/file1_b.fastq',
'/path/to/prefix2_b.fastq']}
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
error = ("Error creating artifact. Offending files:\nraw_forward_seqs:"
" The provided files do not match the run prefix values in "
"the prep information: file1.fastq\n"
"raw_barcodes: The provided files do not match the run "
"prefix values in the prep information: file1_b.fastq")
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertItemsEqual(obs_error.split('\n'), error.split('\n'))
# A required filepath type is missing
files = {'raw_forward_seqs': ['/path/to/prefix1.fastq',
'/path/to/prefix2.fastq'],
'raw_reverse_seqs': ['/path/to/prefix1_rev.fastq',
'/path/to/prefix2_rev.fastq']}
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertEqual(obs_error,
"Missing required filepath type(s): raw_barcodes")
# No run prefix and more than 1 lane
prep_info = {"1.SKB2.640194": {"not_a_run_prefix": "prefix1"},
"1.SKM4.640180": {"not_a_run_prefix": "prefix1"},
"1.SKB3.640195": {"not_a_run_prefix": "prefix2"}}
files = {'raw_forward_seqs': ['/path/to/prefix1.fastq',
'/path/to/prefix2.fastq'],
'raw_barcodes': ['/path/to/prefix1_b.fastq',
'/path/to/prefix2_b.fastq']}
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
error = ("Error creating artifact. Offending files:\nraw_forward_seqs:"
" Only one file per type is allowed. Please provide the "
"column 'run_prefix' if you need more than one file per "
"type: prefix1.fastq, prefix2.fastq\n"
"raw_barcodes: Only one file per type is allowed. Please "
"provide the column 'run_prefix' if you need more than one "
"file per type: prefix1_b.fastq, prefix2_b.fastq")
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertItemsEqual(obs_error.split('\n'), error.split('\n'))
def test_validate_multiple_error(self):
# Filepath type not supported
prep_info = {
"1.SKB2.640194": {
"run_prefix": "prefix1"
},
"1.SKM4.640180": {
"run_prefix": "prefix1"
},
"1.SKB3.640195": {
"run_prefix": "prefix2"
}
}
files = {'Unknown': ['/path/to/file1.fastq']}
atype = "FASTQ"
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertEqual(
obs_error, "Filepath type(s) Unknown not supported by artifact "
"type FASTQ. Supported filepath types: raw_barcodes, "
"raw_forward_seqs, raw_reverse_seqs")
# Number of provided files != Num run prefix values
files = {
'raw_forward_seqs': ['/path/to/file1.fastq'],
'raw_barcodes': [
'/path/to/file1_b.fastq', '/path/to/file2_b.fastq',
'/path/to/file3_b.fastq'
]
}
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
error = ("Error creating artifact. Offending files:\nraw_forward_seqs:"
" The number of provided files (1) doesn't match the number "
"of run prefix values in the prep info (2): file1.fastq\n"
"raw_barcodes: The number of provided files (3) doesn't "
"match the number of run prefix values in the prep info (2): "
"file1_b.fastq, file2_b.fastq, file3_b.fastq")
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertItemsEqual(obs_error.split('\n'), error.split('\n'))
# File doesn't match any run prefix
files = {
'raw_forward_seqs':
['/path/to/file1.fastq', '/path/to/prefix2.fastq'],
'raw_barcodes':
['/path/to/file1_b.fastq', '/path/to/prefix2_b.fastq']
}
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
error = ("Error creating artifact. Offending files:\nraw_forward_seqs:"
" The provided files do not match the run prefix values in "
"the prep information: file1.fastq\n"
"raw_barcodes: The provided files do not match the run "
"prefix values in the prep information: file1_b.fastq")
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertItemsEqual(obs_error.split('\n'), error.split('\n'))
# A required filepath type is missing
files = {
'raw_forward_seqs':
['/path/to/prefix1.fastq', '/path/to/prefix2.fastq'],
'raw_reverse_seqs':
['/path/to/prefix1_rev.fastq', '/path/to/prefix2_rev.fastq']
}
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertEqual(obs_error,
"Missing required filepath type(s): raw_barcodes")
# No run prefix and more than 1 lane
prep_info = {
"1.SKB2.640194": {
"not_a_run_prefix": "prefix1"
},
"1.SKM4.640180": {
"not_a_run_prefix": "prefix1"
},
"1.SKB3.640195": {
"not_a_run_prefix": "prefix2"
}
}
files = {
'raw_forward_seqs':
['/path/to/prefix1.fastq', '/path/to/prefix2.fastq'],
'raw_barcodes':
['/path/to/prefix1_b.fastq', '/path/to/prefix2_b.fastq']
}
job_id = self._create_template_and_job(prep_info, files, atype)
obs_success, obs_ainfo, obs_error = _validate_multiple(
self.qclient, job_id, prep_info, files, atype)
error = ("Error creating artifact. Offending files:\nraw_forward_seqs:"
" Only one file per type is allowed. Please provide the "
"column 'run_prefix' if you need more than one file per "
"type: prefix1.fastq, prefix2.fastq\n"
"raw_barcodes: Only one file per type is allowed. Please "
"provide the column 'run_prefix' if you need more than one "
"file per type: prefix1_b.fastq, prefix2_b.fastq")
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertItemsEqual(obs_error.split('\n'), error.split('\n'))