def test_cmscan_task_multithreaded(self, tmpdir, datadir):
with tmpdir.as_cwd():
transcript = datadir('rnaseP-bsu.fa')
cm = datadir('rnaseP-eubact.c.cm')
out_single = str(tmpdir.join('single'))
out_multi = str(tmpdir.join('multi'))
for n_threads in (2, 3, 4, 5):
db_task = CMPressTask().task(cm, params=self.cmpress_cfg)
aln_tasks_single = CMScanTask().task(transcript,
out_single,
cm,
cutoff=1.0,
n_threads=1)
aln_tasks_multi = CMScanTask().task(transcript,
out_multi,
cm,
cutoff=1.0,
n_threads=n_threads)
run_tasks([db_task, aln_tasks_single], ['run'])
run_task(aln_tasks_multi)
alns_single = pd.concat(InfernalParser(out_single))
alns_multi = pd.concat(InfernalParser(out_multi))
assert all(alns_single['e_value'].sort_values() == \
alns_multi['e_value'].sort_values())
def test_hmmscan_task_multithreaded(self, tmpdir, datadir):
with tmpdir.as_cwd():
prot = datadir('20aa-alitest.fa')
hmm = datadir('20aa.hmm')
out_single = str(tmpdir.join('out-single'))
out_multi = str(tmpdir.join('out-multi'))
for n_threads in (2, 3, 4, 5):
db_task = HMMPressTask().task(hmm, params=self.hmmpress_cfg)
aln_task_single = HMMScanTask().task(prot,
out_single,
hmm,
cutoff=1.0,
n_threads=1)
aln_task_multi = HMMScanTask().task(prot,
out_multi,
hmm,
cutoff=1.0,
n_threads=n_threads)
run_tasks([db_task, aln_task_single], ['run'])
run_tasks([aln_task_multi], ['run'])
print(os.listdir(), file=sys.stderr)
print(open(out_single).read())
alns_single = pd.concat(HMMerParser(out_single))
alns_multi = pd.concat(HMMerParser(out_multi))
assert all(alns_single['domain_i_evalue'].sort_values().reset_index(drop=True) == \
alns_multi['domain_i_evalue'].sort_values().reset_index(drop=True))
def test_hmmpress_task_existing(self, tmpdir, datadir):
with tmpdir.as_cwd():
tf = datadir('test-profile.hmm')
for ext in self.extensions:
touch(tf + ext)
task = HMMPressTask().task(tf)
run_tasks([task], ['run'])
print(os.listdir(), file=sys.stderr)
print(task, file=sys.stderr)
status = check_status(task)
assert status.status == 'up-to-date'
def test_cmpress_task(self, tmpdir, datadir):
with tmpdir.as_cwd():
tf = datadir('test-covariance-model.cm')
task = CMPressTask().task(tf, params=self.cmpress_cfg)
run_tasks([task], ['run'])
status = check_status(task)
print(os.listdir(), file=sys.stderr)
for ext in self.extensions:
assert os.path.isfile(tf + ext)
assert status.status == 'up-to-date'
def get_prices(data):
marketId_entries, urls = data
dbMgr.persist_marketId(marketId_entries)
concurrent_tasks = 10
tasks = []
global counter
counter += 1
if counter % interval_historcial == 0:
counter = 0
def persist_intraday(data):
d = defer.maybeDeferred(dbMgr.persist_intraday_data, data)
return d
def persist_historical(data):
d = defer.maybeDeferred(dbMgr.persist_historical_data, data)
return d
for intraday_url, historical_url in urls:
tasks.append((download_price_data, intraday_url, persist_intraday))
if counter == 0:
tasks.append((download_price_data, historical_url, persist_historical))
defer_list = run_tasks(tasks, concurrent_tasks)
return defer_list
def test_hmmscan_task(self, tmpdir, datadir):
with tmpdir.as_cwd():
prot = datadir('test-protein.fa')
hmm = datadir('test-profile.hmm')
out = str(tmpdir.join('test.out'))
db_task = HMMPressTask().task(hmm, params=self.hmmpress_cfg)
aln_task = HMMScanTask().task(prot, out, hmm,
cutoff=1.0, n_threads=1)
run_tasks([db_task, aln_task], ['run'])
print(os.listdir(), file=sys.stderr)
aln = open(out).read()
print(aln)
assert aln.count('accession') == 2
assert 'i-Evalue' in aln
def test_longorfs_task(self, tmpdir, datadir):
with tmpdir.as_cwd():
transcript = datadir('test-transcript.fa')
exp_orf = datadir('test-transcript-orf.pep')
task = TransDecoderLongOrfsTask().task(transcript,
params=self.longorfs_cfg)
run_tasks([task], ['run'])
output_dir = transcript + '.transdecoder_dir'
exp_pep = open(exp_orf).read()
pep_fn = os.path.join(output_dir, 'longest_orfs.pep')
assert os.path.isfile(pep_fn)
pep = open(pep_fn).read()
assert exp_pep in pep
def test_cmscan_task(self, tmpdir, datadir):
with tmpdir.as_cwd():
transcript = datadir('test-transcript.fa')
cm = datadir('test-covariance-model.cm')
out = str(tmpdir.join('test.out'))
db_task = CMPressTask().task(cm, params=self.cmpress_cfg)
aln_task = CMScanTask().task(transcript,
out,
cm,
cutoff=1.0,
n_threads=1)
run_tasks([db_task, aln_task], ['run'])
print(os.listdir(), file=sys.stderr)
aln = ''.join(open(out).readlines())
print(aln, file=sys.stderr)
# TODO: better correctness check
assert aln.count('accession') == 2
assert 'E-value' in aln
def test_predict_task(self, tmpdir, datadir):
with tmpdir.as_cwd():
transcript = datadir('pom.50.fa')
pfam = datadir('test-protein-x-pfam-a.tbl')
orf_task = TransDecoderLongOrfsTask().task(
transcript, params=self.longorfs_cfg)
pred_task = TransDecoderPredictTask().task(transcript,
pfam,
params=self.predict_cfg)
run_tasks([orf_task, pred_task], ['run'])
pprint(tmpdir.listdir())
for ext in self.extensions:
fn = transcript + '.transdecoder' + ext
assert os.path.isfile(fn)
contents = open(fn).read()
if ext == '.gff3':
assert 'mRNA' in contents
assert 'gene' in contents
assert 'CDS' in contents
assert 'three_prime_UTR' in contents
assert 'exon' in contents