def test_upsert_tuple(self):
"""
Checks to make sure that the client accepts both tuples and lists.
:return:
"""
connection = celldb.connect(URL)
celldb.upsert_samples(connection, ('sample_tuple', ),
('feature_tuple', ), ((0.1, ), ))
def test_upsert_samples(self):
cursor = celldb.connect(URL)
sample_ids = ["sample_{}".format(x) for x in range(20)]
feature_ids = ["feature_{}".format(x) for x in range(10)]
vectors = [[random.random() for x in range(len(feature_ids))]
for x in range(len(sample_ids))]
celldb.upsert_samples(cursor, sample_ids, feature_ids, vectors)
assert set(celldb.list_samples(cursor)) == set(sample_ids)
assert set(celldb.list_features(cursor)) == set(feature_ids)
def test_upsert_100000_wide(self):
cursor = celldb.connect(URL)
per_sample = 200 # set the minimum threshold of time per sample
n_samples = 1
n_features = 100000
sampleIds, featureIds, vectors = _random_dataset(n_samples, n_features)
start = time.time()
celldb.upsert_samples(cursor, sampleIds, featureIds, vectors)
end = time.time()
assert (end - start) / float(n_samples) < per_sample
def test_single_feature_matrix(self):
"""
Tests to make sure that making a single feature is upserted as
expected and that retrieving the resulting matrix doesn't error.
:return:
"""
connection = celldb.connect(URL)
sampleIds, featureIds, vectors = _random_dataset(1, 4)
celldb.upsert_samples(connection, sampleIds, featureIds, vectors)
feature_ids = celldb.list_features(connection)
assert set(feature_ids) == set(featureIds)
_drop_tables(connection)
def test_matrix(self):
cursor = celldb.connect(URL)
sampleIds, featureIds, vectors = _random_dataset(4, 4)
celldb.upsert_samples(cursor, sampleIds, featureIds, vectors)
matrix = celldb.matrix(cursor, sampleIds, featureIds)
for k, row in enumerate(matrix):
assert row[0] == sampleIds[k]
assert len(row) == len(featureIds) + 1
for i, value in enumerate(row[1:]):
assert value == vectors[k][i]
assert len(matrix) == len(sampleIds)
_drop_tables(cursor)
def test_sparse_matrix(self):
cursor = celldb.connect(URL)
sample_ids, feature_ids, vectors = _random_dataset(4, 4)
celldb.upsert_samples(cursor, sample_ids, feature_ids, vectors)
matrix = celldb.sparse_matrix(cursor, sample_ids, feature_ids)
# The sparse matrix has the list of samples and features at the top
# level for convenience. These become indices into the `values` map.
assert len(matrix['sample_ids']) == len(sample_ids)
assert len(matrix['feature_ids']) == len(feature_ids)
for k, v in enumerate(matrix['values']):
assert vectors[int(k)][int(v)] == vectors[int(k)][int(v)]
_drop_tables(cursor)
def test_large_matrix(self):
"""
Tests to make sure that when request a large matrix we get satisfactory
results.
:return:
"""
connection = celldb.connect(URL)
# The amount of time in seconds per point we're willing to accept
per_point = 0.01
n_samples = 10000
n_features = 40
points = float(n_samples * n_features)
sampleIds, featureIds, vectors = _random_dataset(10000, 40)
celldb.upsert_samples(connection, sampleIds, featureIds, vectors)
start = time.time()
matrix = celldb.matrix(connection, sampleIds, featureIds)
assert len(matrix) == len(sampleIds)
end = time.time()
assert (end - start) / points < per_point
def test_singleton_matrix(self):
"""
Tests to make sure a single sample doesn't throw an exception and
returns the expected results.
:return:
"""
connection = celldb.connect(URL)
sampleIds, featureIds, vectors = _random_dataset(1, 4)
celldb.upsert_samples(connection, sampleIds, featureIds, vectors)
sample_ids = celldb.list_samples(connection)
assert list(sample_ids)[0] == sampleIds[0]
feature_ids = celldb.list_features(connection)
assert len(featureIds) == len(list(feature_ids))
matrix = celldb.matrix(connection, sampleIds, featureIds)
for k, row in enumerate(matrix):
assert row[0] == sampleIds[k]
# The row has the sample_id in the first position
assert len(row) == len(featureIds) + 1
for i, value in enumerate(row[1:]):
assert value == vectors[k][i]
_drop_tables(connection)
def test_list_many(self):
"""
Makes sure both upsert and list happen in appropriate time
:return:
"""
connection = celldb.connect(URL)
_drop_tables(connection)
list_time = 30
per_sample = 1
n_samples = 10000
n_features = 10
sample_ids, feature_ids, vectors = _random_dataset(
n_samples, n_features)
start = time.time()
celldb.upsert_samples(connection, sample_ids, feature_ids, vectors)
end = time.time()
assert (end - start) / float(n_samples) < per_sample
start = time.time()
sample_list = list(celldb.list_samples(connection))
end = time.time()
assert (end - start) < list_time
assert len(sample_list) == len(sample_ids)