def test_validate_demux_file_error(self):
demux_fp, _, out_dir = self._generate_files({'s1': 's1', 's2': 's2'})
# Run prefix not provided and demux samples do not match
prep_info = {
"1.SKB2.640194": {
"not_a_run_prefix": "prefix1"
},
"1.SKM4.640180": {
"not_a_run_prefix": "prefix2"
},
"1.SKB3.640195": {
"not_a_run_prefix": "prefix3"
}
}
files = {'preprocessed_demux': [demux_fp]}
job_id = self._create_template_and_job(prep_info, files,
"Demultiplexed")
obs_success, obs_ainfo, obs_error = _validate_demux_file(
self.qclient, job_id, prep_info, out_dir, demux_fp)
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertEqual(
obs_error, 'The sample ids in the demultiplexed files do not '
'match the ones in the prep information. Please, '
'provide the column "run_prefix" in the prep '
'information to map the existing sample ids to the '
'prep information sample ids.')
# Incorrect run prefix column
prep_info = {
"1.SKB2.640194": {
"run_prefix": "prefix1"
},
"1.SKM4.640180": {
"run_prefix": "prefix2"
},
"1.SKB3.640195": {
"run_prefix": "prefix3"
}
}
job_id = self._create_template_and_job(prep_info, files,
"Demultiplexed")
obs_success, obs_ainfo, obs_error = _validate_demux_file(
self.qclient, job_id, prep_info, out_dir, demux_fp)
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertEqual(
obs_error, 'The sample ids in the "run_prefix" columns from the '
'prep information do not match the ones in the demux '
'file. Please, correct the column "run_prefix" in '
'the prep information to map the existing sample ids '
'to the prep information sample ids.')
def test_validate_demux_file_infer(self):
demux_fp, _, out_dir = self._generate_files({'s1': 'SKB2.640194',
's2': 'SKM4.640180'})
prep_info = {"1.SKB2.640194": {"not_a_run_prefix": "s1"},
"1.SKM4.640180": {"not_a_run_prefix": "s2"},
"1.SKB3.640195": {"not_a_run_prefix": "s3"},
"1.SKB6.640176": {"not_a_run_prefix": "s4"}}
files = {'preprocessed_demux': [demux_fp]}
job_id = self._create_template_and_job(
prep_info, files, "Demultiplexed")
obs_success, obs_ainfo, obs_error = _validate_demux_file(
self.qclient, job_id, prep_info, out_dir, demux_fp)
self.assertTrue(obs_success)
name = splitext(basename(demux_fp))[0]
exp_fastq_fp = join(out_dir, "%s.fastq" % name)
exp_fasta_fp = join(out_dir, "%s.fasta" % name)
exp_demux_fp = join(out_dir, basename(demux_fp))
filepaths = [
(exp_fastq_fp, 'preprocessed_fastq'),
(exp_fasta_fp, 'preprocessed_fasta'),
(exp_demux_fp, 'preprocessed_demux')]
exp = [ArtifactInfo(None, "Demultiplexed", filepaths)]
self.assertEqual(obs_ainfo, exp)
self.assertEqual(obs_error, "")
with File(exp_demux_fp) as f:
self.assertItemsEqual(f.keys(), ["1.SKB2.640194", "1.SKM4.640180"])
def test_validate_demux_file(self):
demux_fp, _, out_dir = self._generate_files({'s1': 's1', 's2': 's2'})
prep_info = {
"1.SKB2.640194": {
"run_prefix": "s1"
},
"1.SKM4.640180": {
"run_prefix": "s2"
},
"1.SKB3.640195": {
"run_prefix": "s3"
},
"1.SKB6.640176": {
"run_prefix": "s4"
}
}
files = {'preprocessed_demux': demux_fp}
job_id = self._create_template_and_job(prep_info, files,
"Demultiplexed")
obs_success, obs_ainfo, obs_error = _validate_demux_file(
self.qclient, job_id, prep_info, out_dir, demux_fp)
self.assertTrue(obs_success)
name = splitext(basename(demux_fp))[0]
exp_fastq_fp = join(out_dir, "%s.fastq" % name)
exp_fasta_fp = join(out_dir, "%s.fasta" % name)
exp_demux_fp = join(out_dir, basename(demux_fp))
filepaths = [(exp_fastq_fp, 'preprocessed_fastq'),
(exp_fasta_fp, 'preprocessed_fasta'),
(exp_demux_fp, 'preprocessed_demux')]
exp = [ArtifactInfo(None, "Demultiplexed", filepaths)]
self.assertEqual(obs_ainfo, exp)
self.assertEqual(obs_error, "")
with File(exp_demux_fp) as f:
self.assertItemsEqual(f.keys(), ["1.SKB2.640194", "1.SKM4.640180"])
def test_validate_demux_file_error(self):
demux_fp, _, out_dir = self._generate_files({'s1': 's1', 's2': 's2'})
# Run prefix not provided and demux samples do not match
prep_info = {"1.SKB2.640194": {"not_a_run_prefix": "prefix1"},
"1.SKM4.640180": {"not_a_run_prefix": "prefix2"},
"1.SKB3.640195": {"not_a_run_prefix": "prefix3"}}
files = {'preprocessed_demux': [demux_fp]}
job_id = self._create_template_and_job(
prep_info, files, "Demultiplexed")
obs_success, obs_ainfo, obs_error = _validate_demux_file(
self.qclient, job_id, prep_info, out_dir, demux_fp)
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertEqual(obs_error,
'The sample ids in the demultiplexed files do not '
'match the ones in the prep information. Please, '
'provide the column "run_prefix" in the prep '
'information to map the existing sample ids to the '
'prep information sample ids.')
# Incorrect run prefix column
prep_info = {"1.SKB2.640194": {"run_prefix": "prefix1"},
"1.SKM4.640180": {"run_prefix": "prefix2"},
"1.SKB3.640195": {"run_prefix": "prefix3"}}
job_id = self._create_template_and_job(
prep_info, files, "Demultiplexed")
obs_success, obs_ainfo, obs_error = _validate_demux_file(
self.qclient, job_id, prep_info, out_dir, demux_fp)
self.assertFalse(obs_success)
self.assertIsNone(obs_ainfo)
self.assertEqual(obs_error,
'The sample ids in the "run_prefix" columns from the '
'prep information do not match the ones in the demux '
'file. Please, correct the column "run_prefix" in '
'the prep information to map the existing sample ids '
'to the prep information sample ids.')