def test_dask_local_name(self):
self.assertEqual(MPControl.name(), self.client_name)
def test_mp_sync(self):
self.assertTrue(MPControl.sync_processes())
def setUpClass(cls):
cls.tempdir = tempfile.mkdtemp()
MPControl.shutdown()
MPControl.set_multiprocess_engine(cls.name)
MPControl.connect(local_dir=cls.tempdir, n_workers=0)
def tearDownClass(cls):
MPControl.shutdown()
MPControl.set_multiprocess_engine("local")
MPControl.connect()
def start_mpcontrol_dask(n_cores=N_CORES):
MPControl.set_multiprocess_engine("dask-cluster")
MPControl.client.use_default_configuration("rusty_ccb", n_jobs=2)
MPControl.client.add_worker_conda(
"source /mnt/ceph/users/sysbio/.anaconda3/bin/activate inferelator")
MPControl.connect()
def test_dask_cluster_map(self):
with self.assertRaises(NotImplementedError):
MPControl.map(math_function, *self.map_test_data)
def setUpClass(cls):
MPControl.shutdown()
MPControl.set_multiprocess_engine(cls.name)
MPControl.connect()
def setUpClass(cls):
cls.temp_dir = tempfile.mkdtemp()
cls.server = kvsstcp.KVSServer("", 0)
MPControl.shutdown()
MPControl.set_multiprocess_engine(cls.name)
MPControl.connect(host=cls.server.cinfo[0], port=cls.server.cinfo[1])
def test_kvs_map_distribute(self):
test_result = MPControl.map(math_function,
*self.map_test_data,
tell_children=True)
self.assertListEqual(test_result, self.map_test_expect)
def test_bad_engine_II(self):
with self.assertRaises(ValueError):
MPControl.set_multiprocess_engine(unittest.TestCase)
def test_cant_change(self):
with self.assertRaises(RuntimeError):
MPControl.set_multiprocess_engine("local")
def test_bad_engine(self):
with self.assertRaises(ValueError):
MPControl.set_multiprocess_engine("V8")
def test_name(self):
with self.assertRaises(NameError):
MPControl.name()
def test_sync(self):
with self.assertRaises(RuntimeError):
MPControl.sync_processes()
def test_dask_local_sync(self):
self.assertTrue(MPControl.sync_processes())
def test_kvs_map_by_file(self):
test_result = MPControl.map(math_function,
*self.map_test_data,
tell_children=True,
tmp_file_path=self.temp_dir)
self.assertListEqual(test_result, self.map_test_expect)
def test_dask_cluster_name(self):
self.assertEqual(MPControl.name(), self.client_name)
def test_kvs_sync(self):
self.assertEqual(MPControl.sync_processes(), None)
def test_dask_cluster_sync(self):
self.assertTrue(MPControl.sync_processes())
def test_mp_name(self):
self.assertEqual(MPControl.name(), self.name)
def setUpClass(cls):
MPControl.shutdown()
MPControl.client = None
def test_mp_map(self):
test_result = MPControl.map(math_function, *self.map_test_data)
self.assertListEqual(test_result, self.map_test_expect)
def test_map(self):
with self.assertRaises(RuntimeError):
MPControl.map(math_function, *self.map_test_data)
DATA_FILE = "YPD_LOCAL_OPT_PV_DEXFORM.h5ad"
CONDA_ACTIVATE_PATH = '~/.local/anaconda3/bin/activate'
EXPRESSION_MATRIX_METADATA = ['Genotype', 'Genotype_Group', 'Replicate', 'Condition', 'tenXBarcode']
YEASTRACT_PRIOR = "YEASTRACT_20190713_BOTH.tsv"
TF_NAMES = "tf_names_gold_standard.txt"
YEASTRACT_TF_NAMES = "tf_names_yeastract.txt"
INPUT_DIR = '/mnt/ceph/users/cjackson/inferelator/data/yeast'
OUTPUT_PATH = '/mnt/ceph/users/cjackson/jackson_2019_inferelator_v050/'
if __name__ == '__main__':
MPControl.set_multiprocess_engine("dask-cluster")
MPControl.client.use_default_configuration("rusty_preempt")
MPControl.client.set_job_size_params(n_jobs=1)
MPControl.client.add_worker_conda("source ~/.local/anaconda3/bin/activate inferelator")
MPControl.connect()
wkf = inferelator_workflow("stars", VelocityWorkflow)
wkf.set_file_paths(input_dir=INPUT_DIR,
output_dir=os.path.join(OUTPUT_PATH, "only_v"),
gold_standard_file='gold_standard.tsv',
priors_file=YEASTRACT_PRIOR,
tf_names_file=YEASTRACT_TF_NAMES)
wkf.set_expression_file(h5ad=DATA_FILE, h5_layer="smooth_count")
wkf.set_velocity_parameters(velocity_file_name=DATA_FILE, velocity_file_type="h5ad", velocity_file_layer="pv")
wkf.set_crossvalidation_parameters(split_gold_standard_for_crossvalidation=True,
cv_split_ratio=0.5)
