def gen_or_read_zebrafish_data():
# generate data if needed
if not os.path.exists(test_zebrafish_data_path):
print("generating test data...")
gen_zebrafish_test_data()
print("reading test data...")
# To-do: use a fixture in future
adata = dyn.read_h5ad(test_zebrafish_data_path)
return adata
def color_set(data, gene):
gene_idx = np.where(data.var_names == gene)[0][0]
c = np.squeeze(np.asarray(data.X[:, gene_idx].todense()))
return c
dataset = sys.argv[1]
embs = ['X_pca', 'X_umap']
#3 is the stationary distribution, it will used later
plots = ['FN1', 'VIM', 3]
P = np.loadtxt(dataset + '_P.npy')
Idx = np.load(dataset + '_idx.npy', allow_pickle=True)
adata = dyn.read_h5ad(dataset + '_processed.h5ad')
kmc = dyn.tl.KernelMarkovChain(P=P, Idx=Idx)
X = adata.layers['M_s'][:, adata.var['use_for_velocity']]
V = adata.layers['velocity_S'][:, adata.var['use_for_velocity']]
sd = kmc.compute_stationary_distribution()
for plot in plots:
for emb in embs:
X_emb = adata.obsm[emb]
Uc = kmc.compute_density_corrected_drift(X_emb, normalize_vector=True)
U_grid, X_grid = dyn.tl.smoothen_drift_on_grid(X_emb[:, :2],
Uc[:, :2],
30,
k=50,
def read_test_spatial_genomics_data():
return dyn.read_h5ad(test_spatial_genomics_path)
filename = sys.argv[1]
dataset_name = filename.split("/")[-1].split(".")[0]
#file_extension=filename.split(".")[-1]
file_extension = 'h5ad'
processed_output_path = dataset_name + '_processed.' + file_extension
P_output_path = dataset_name + '_P.' + 'npy'
idx_output_path = dataset_name + '_idx.' + 'npy'
print()
print('Loading Data')
print()
if filename.endswith('.h5ad'):
adata = dyn.read_h5ad(filename)
elif filename.endswith('.loom'):
adata = dyn.read_loom(filename)
else:
print('file type not currently supported')
print()
print('Initial Preprocessing')
print()
dyn.pp.recipe_monocle(adata, n_top_genes=2000)
print()
print('Computing Dynamics')
print()
dyn.tl.dynamics(adata, mode='moment')
def test_space_data():
adata = dyn.read_h5ad("allstage_splice.h5ad")
# dyn.tl.cluster_community(adata, method="infomap")
assert np.all(adata.obs["louvain"] != -1)
assert np.all(adata.obs["leiden"] != -1)
assert np.all(adata.obs["infomap"] != -1)
# dyn.pl.louvain(adata, basis="pca")
# dyn.pl.leiden(adata, basis="pca")
# dyn.pl.infomap(adata, basis="pca")
def test_simple_cluster_field(adata):
dyn.tl.cluster_field(adata, method="louvain")
dyn.tl.cluster_field(adata, method="leiden")
if __name__ == "__main__":
# generate data if needed
if not os.path.exists(test_data_path):
print("generating test data...")
gen_test_data()
print("reading test data...")
# To-do: use a fixture in future
adata = dyn.read_h5ad(test_data_path)
print(adata)
# select a subset of adata for testing
print("tests begin...")
######### testing begins here #########
test_simple_cluster_community_adata(adata)
# test_simple_cluster_field(adata)