def test_cmscan_task_multithreaded(self):
with TemporaryDirectory() as td:
with Move(td):
with TestData('rnaseP-bsu.fa', td) as transcript, \
TestData('rnaseP-eubact.c.cm', td) as cm, \
TemporaryFile(td) as out_single,\
TemporaryFile(td) as out_multi:
for n_threads in (2,3,4,5):
db_task = tasks.get_cmpress_task(cm, self.cmpress_cfg)
aln_task_single = tasks.get_cmscan_task(transcript, out_single,
cm, 1.0, 1,
self.cmscan_cfg)
aln_task_multi = tasks.get_cmscan_task(transcript, out_multi,
cm, 1.0,
n_threads,
self.cmscan_cfg)
run_tasks([db_task, aln_task_single], ['run'])
run_task(aln_task_multi)
alns_single = pd.concat(cmscan_to_df_iter(out_single))
alns_multi = pd.concat(cmscan_to_df_iter(out_multi))
self.assertTrue(all(alns_single['e_value'].sort_values() == \
alns_multi['e_value'].sort_values()))
def test_cmscan_task_multithreaded(self):
with TemporaryDirectory() as td:
with Move(td):
with TestData('rnaseP-bsu.fa', td) as transcript, \
TestData('rnaseP-eubact.c.cm', td) as cm, \
TemporaryFile(td) as out_single,\
TemporaryFile(td) as out_multi:
for n_threads in (2, 3, 4, 5):
db_task = tasks.get_cmpress_task(cm, self.cmpress_cfg)
aln_task_single = tasks.get_cmscan_task(
transcript, out_single, cm, 1.0, 1,
self.cmscan_cfg)
aln_task_multi = tasks.get_cmscan_task(
transcript, out_multi, cm, 1.0, n_threads,
self.cmscan_cfg)
run_tasks([db_task, aln_task_single], ['run'])
run_task(aln_task_multi)
alns_single = pd.concat(cmscan_to_df_iter(out_single))
alns_multi = pd.concat(cmscan_to_df_iter(out_multi))
self.assertTrue(all(alns_single['e_value'].sort_values() == \
alns_multi['e_value'].sort_values()))
def test_cmscan_task(self):
with TemporaryDirectory() as td:
with Move(td):
with TestData('test-transcript.fa', td) as transcript, \
TestData('test-covariance-model.cm', td) as cm, \
TemporaryFile(td) as out:
db_task = tasks.get_cmpress_task(cm, self.cmpress_cfg)
aln_task = tasks.get_cmscan_task(transcript, out, cm, 1.0, 1,
self.cmscan_cfg)
run_tasks([db_task, aln_task], ['run'])
print(os.listdir(td), file=sys.stderr)
aln = ''.join(open(out).readlines())
print(aln, file=sys.stderr)
# TODO: better correctness check
self.assertEquals(aln.count('accession'), 2)
self.assertIn('E-value', aln)
def test_cmscan_task(self):
with TemporaryDirectory() as td:
with Move(td):
with TestData('test-transcript.fa', td) as transcript, \
TestData('test-covariance-model.cm', td) as cm, \
TemporaryFile(td) as out:
db_task = tasks.get_cmpress_task(cm, self.cmpress_cfg)
aln_task = tasks.get_cmscan_task(transcript, out, cm, 1.0, 1,
self.cmscan_cfg)
run_tasks([db_task, aln_task], ['run'])
print(os.listdir(td), file=sys.stderr)
aln = ''.join(open(out).readlines())
print(aln, file=sys.stderr)
# TODO: better correctness check
self.assertEquals(aln.count('accession'), 2)
self.assertIn('E-value', aln)