def test_lastal_task_multithreaded(tmpdir, datadir):
with tmpdir.as_cwd():
for n_threads in (3,4,5):
prot = datadir('test-protein.fa')
tr = datadir('pom.50.fa')
out_single = tmpdir.join('out-single').strpath
out_multi = tmpdir.join('out-multi').strpath
db_task = lastdb_task(prot, prot)
aln_task_single = lastal_task(tr, prot, out_single,
translate=True,
cutoff=None)
aln_task_multi = lastal_task(tr, prot, out_multi,
translate=True,
cutoff=None,
n_threads=n_threads)
run_tasks([db_task, aln_task_multi, aln_task_single],
['run'])
alns_single = MafParser(out_single).read()
alns_multi = MafParser(out_multi).read()
assert all(alns_single['E'].sort_values() == \
alns_multi['E'].sort_values())
def test_crbl(tmpdir, datadir):
query = datadir('sacPom.cdna.fa')
database = datadir('sacPom.pep.fa')
out_fn = str(tmpdir.join('test.csv'))
args = ['crbl', '-q', query, '-d', database, '-o', out_fn]
status, out, err = runscript('shmlast', args, directory=str(tmpdir))
assert status == 0
assert tmpdir.ensure(out_fn)
def test_reciprocal_best_hits(datadir):
query_df = pd.read_csv(datadir('query.maf.csv'))
db_df = pd.read_csv(datadir('db.maf.csv'))
expected_df = pd.read_csv(datadir('reciprocals.expected.csv'))
bh = BestHits()
results_df = bh.reciprocal_best_hits(query_df, db_df, inplace=False)
assert check_df_equals(results_df, expected_df)
def test_rbl(tmpdir, datadir):
query = datadir('sacPom.cdna.fa')
database = datadir('sacPom.pep.fa')
print(query, database, tmpdir)
args = ['rbl', '-q', query, '-d', database]
runscript('shmlast', args, directory=str(tmpdir))
assert tmpdir.ensure('sacPom.cdna.fa.x.sacPom.pep.fa.rbl.csv')
def test_reciprocal_best_hits(datadir):
query_df = pd.read_csv(datadir('query.maf.csv'))
db_df = pd.read_csv(datadir('db.maf.csv'))
expected_df = pd.read_csv(datadir('reciprocals.expected.csv'))
bh = BestHits()
results_df = bh.reciprocal_best_hits(query_df, db_df, inplace=False)
assert check_df_equals(results_df, expected_df)
def test_crbl_tasks_empty(tmpdir, datadir):
with tmpdir.as_cwd():
input_fa = datadir('pom.single.fa')
pep_fa = datadir('odb_subset.fa')
results_fn = tmpdir.join('result.csv').strpath
crbl = CRBL(input_fa, pep_fa, results_fn)
result = run_tasks([tsk for tsk in crbl.tasks()], ['run'])
assert result == 0
def test_besthits_non_inplace(datadir):
'''Test BestHits.best_hits with inplace=False
'''
input_df = pd.read_csv(datadir('query.maf.csv'))
expected_df = pd.read_csv(datadir('besthits.expected.csv'))
bh = BestHits()
results_df = bh.best_hits(input_df, inplace=False)
assert check_df_equals(expected_df, results_df)
def test_besthits_non_inplace(datadir):
'''Test BestHits.best_hits with inplace=False
'''
input_df = pd.read_csv(datadir('query.maf.csv'))
expected_df = pd.read_csv(datadir('besthits.expected.csv'))
bh = BestHits()
results_df = bh.best_hits(input_df, inplace=False)
assert check_df_equals(expected_df, results_df)
def test_crbl_tasks_empty(tmpdir, datadir):
with tmpdir.as_cwd():
input_fa = datadir('pom.single.fa')
pep_fa = datadir('odb_subset.fa')
results_fn = tmpdir.join('result.csv').strpath
crbl = CRBL(input_fa,
pep_fa,
results_fn)
result = run_tasks([tsk for tsk in crbl.tasks()], ['run'])
assert result == 0
def test_rbl(tmpdir, datadir, n_threads, benchmark):
query = datadir('sacPom.cdna.fa')
database = datadir('sacPom.pep.fa')
args = ['rbl', '--n_threads', str(n_threads), '-q', query, '-d', database]
status, out, err = benchmark.pedantic(runscript,
args=('shmlast', args),
kwargs={'directory': str(tmpdir)},
iterations=1,
rounds=1)
assert status == 0
assert tmpdir.ensure('sacPom.cdna.fa.x.sacPom.pep.fa.rbl.csv')
def test_rbl(tmpdir, datadir, n_threads, benchmark):
query = datadir('sacPom.cdna.fa')
database = datadir('sacPom.pep.fa')
args = ['rbl', '--n_threads', str(n_threads),
'-q', query, '-d', database]
status, out, err = benchmark.pedantic(runscript,
args=('shmlast', args),
kwargs={'directory': str(tmpdir)},
iterations=1,
rounds=1)
assert status == 0
assert tmpdir.ensure('sacPom.cdna.fa.x.sacPom.pep.fa.rbl.csv')
def test_lastal_task_nucl_x_prot(tmpdir, datadir):
with tmpdir.as_cwd():
prot = datadir('test-protein.fa')
tr = datadir('test-transcript.fa')
out = tmpdir.join('test-out').strpath
db_task = lastdb_task(prot, prot)
aln_task = lastal_task(tr, prot, out,
translate=True,
cutoff=None)
run_tasks([db_task, aln_task], ['run'])
aln = ''.join(open(out).readlines())
print(aln, file=sys.stderr)
assert 'SPAC212_RecQ_type_DNA_helicase_PROTEIN' in aln
assert 'SPAC212_RecQ_type_DNA_helicase_TRANSCRIPT' in aln
assert 'lambda' in aln, 'lambda missing, wrong LAST version?'
def lastdb_dir(tmpdir_factory, datadir):
d = tmpdir_factory.mktemp('sacpom_lastdb')
with d.as_cwd():
data = datadir('sacPom.pep.fa')
task = lastdb_task(data, data, prot=True)
result = run_tasks([task], ['run'])
assert result == 0
return d
def test_lastdb_task_existing(tmpdir, datadir):
with tmpdir.as_cwd():
tf = datadir('test-protein.fa')
for ext in LASTDB_EXTENSIONS:
touch(tf + ext)
task = lastdb_task(tf, tf, prot=True)
run_tasks([task], ['run'])
print(task, file=sys.stderr)
status = check_status(task)
assert status.status == 'up-to-date'
def test_lastdb_task_prot(tmpdir, datadir):
with tmpdir.as_cwd():
tf = datadir('test-protein.fa')
task = lastdb_task(tf, tf, prot=True)
run_tasks([task], ['run'])
status = check_status(task)
for ext in LASTDB_EXTENSIONS:
assert os.path.isfile(tf + ext)
assert status.status == 'up-to-date'
def test_lastdb_task_nucl(tmpdir, datadir):
with tmpdir.as_cwd():
tf = datadir('test-transcript.fa')
task = lastdb_task(tf, tf, prot=False)
run_tasks([task], ['run'])
status = check_status(task)
print('PATH:', os.environ['PATH'], file=sys.stderr)
for ext in LASTDB_EXTENSIONS:
assert os.path.isfile(tf + ext)
assert status.status == 'up-to-date'
def test_lastal_task_large(datadir, lastdb_dir, tmpdir_factory, benchmark, n_threads):
with tmpdir_factory.mktemp('THREADS_{0}'.format(n_threads)).as_cwd():
query = datadir('sacPom.cdna.fa')
database = str(lastdb_dir.join('sacPom.pep.fa'))
output = 'out'
aln_task = lastal_task(query, database, output,
translate=True,
cutoff=None,
n_threads=n_threads)
result = benchmark.pedantic(run_tasks,
args=([aln_task], ['run']),
iterations=1,
rounds=1)
assert result == 0
def test_lastal_task_uptodate(tmpdir, datadir):
with tmpdir.as_cwd():
prot = datadir('test-protein.fa')
out = tmpdir.join('test-out').strpath
db_task = lastdb_task(prot, prot)
aln_task = lastal_task(prot, prot, out,
translate=False,
cutoff=None)
# Run it once
run_tasks([db_task, aln_task], ['run'])
# Now run again and check the status
#run_tasks(aln_tasks, ['run'])
print(aln_task)
status = check_status(aln_task, tasks=[aln_task, db_task])
assert status.status == 'up-to-date'
