def test_run_mock_cells():
aname = 'Darmanis_2015_nofetal'
atlas = AtlasFetcher().fetch_atlas(aname, kind='average')
ind = [0, 0, 0, 1, 1, 1, 2, 2, 2, 3, 4, 5]
matrix = atlas[ind]
cell_types = atlas.obs_names.values[ind]
sa = Averages(
aname,
n_pcs=10,
)
sa.fit(matrix)
assert ((cell_types == sa.membership).mean() > 0.9)
def test_arguments():
matrix = pd.DataFrame(
index=['INS', 'GCG', 'PPY'],
columns=['cell1', 'cell2', 'cell3'],
data=[[2302, 123, 0], [0, 5034, 6453], [0, 0, 1]],
)
sa = Averages('Enge_2017',
n_neighbors=2,
threshold_neighborhood=0.8,
n_pcs=2,
distance_metric='correlation',
n_neighbors_out_of_atlas=1)
sa.new_data = matrix
sa._check_init_arguments()
assert (sa is not None)
def test_constructor():
sa = Averages('Enge_2017',
n_neighbors=2,
threshold_neighborhood=0.8,
n_pcs=2,
distance_metric='correlation',
n_neighbors_out_of_atlas=1)
assert (sa is not None)
def test_embed_averages():
aname = 'Darmanis_2015_nofetal'
atlas = AtlasFetcher().fetch_atlas(aname, kind='subsample')
ind = [
0, 2, 5, 8, 10, 15, 20, 25, 28, 30, 35, 38, 40, 45, 50, 60, 70, 75, 80,
90
]
matrix = atlas[ind].copy()
sa = Averages(
aname,
n_features_per_cell_type=2,
n_features_overdispersed=5,
n_pcs=9,
)
sa.fit(matrix)
vs = sa.embed()
assert (vs.shape[1] == 4)
def test_run_within_atlas():
aname = 'Darmanis_2015_nofetal'
atlas = AtlasFetcher().fetch_atlas(aname, kind='subsample')
ind = [
0, 2, 5, 8, 10, 15, 20, 25, 28, 30, 35, 38, 40, 45, 50, 60, 70, 75, 80,
90
]
matrix = atlas[ind]
cell_types = atlas.obs['CellType'].values[ind]
sa = Averages(
aname,
n_features_per_cell_type=2,
n_features_overdispersed=5,
n_pcs=9,
)
sa.fit(matrix)
for i in range(len(cell_types)):
print('{:10s} {:10s}'.format(cell_types[i], sa.membership[i]))
print((cell_types == sa.membership).mean())
assert ((cell_types == sa.membership).mean() >= 0.5)
def test_neighbors_small():
matrix = pd.DataFrame(
index=['INS', 'GCG', 'PPY'],
columns=['cell1', 'cell2', 'cell3'],
data=[[2302, 123, 0], [0, 5034, 6453], [0, 0, 1]],
)
n_pcs = 2
k = 1
threshold = 0.8
sa = Averages('Enge_2017',
n_neighbors=k,
threshold_neighborhood=threshold,
n_pcs=n_pcs,
distance_metric='correlation',
n_cells_per_type=20,
n_neighbors_out_of_atlas=1)
sa.new_data = matrix
sa._check_init_arguments()
sa.fetch_atlas_if_needed()
sa.compute_feature_intersection()
sa._check_feature_intersection()
sa.prepare_feature_selection()
sa.select_features()
sa._check_feature_selection()
sa.merge_atlas_newdata()
sa.compute_pca()
sa.compute_similarity_graph()
neis = sa.neighbors
assert (isinstance(neis, list))
assert (len(neis) == int(np.sum(sa.sizes)))
for nei in neis:
assert (isinstance(nei, list))
def test_merge_small():
matrix = pd.DataFrame(
index=['INS', 'GCG', 'PPY'],
columns=['cell1', 'cell2', 'cell3'],
data=[[2302, 123, 0], [0, 5034, 6453], [0, 0, 1]],
)
n_pcs = 2
k = 1
threshold = 0.8
sa = Averages('Enge_2017',
n_neighbors=k,
threshold_neighborhood=threshold,
n_pcs=n_pcs,
distance_metric='correlation',
n_neighbors_out_of_atlas=1)
sa.new_data = matrix
sa._check_init_arguments()
sa.fetch_atlas_if_needed()
sa.compute_feature_intersection()
sa._check_feature_intersection()
sa.prepare_feature_selection()
sa.select_features()
sa._check_feature_selection()
sa.merge_atlas_newdata()
assert (sa is not None)