update test_basic and datasets

docs/source/index.rst

@@ -1,6 +1,8 @@
 scvelo – stochastic single cell RNA velocity
 ==========================================================
 
+|PyPI| |Docs| |travis|
+
 **scvelo** is a scalable toolkit for estimating and analyzing stochastic RNA velocities in single cells.
 
 RNA velocity is the time derivative of mRNA abundance obtained by distinguishing unspliced (precursor) from spliced
@@ -12,6 +14,15 @@ incorporates intrinsic expression variability.
 
 It is compatible with scanpy_ (`Wolf et al., 2018 <https://doi.org/10.1186/s13059-017-1382-0>`_).
 
+.. |PyPI| image:: https://img.shields.io/pypi/v/scvelo.svg
+    :target: https://pypi.org/project/scvelo
+
+.. |Docs| image:: https://readthedocs.org/projects/scvelo/badge/?version=latest
+   :target: https://scvelo.readthedocs.io
+
+.. |travis| image:: https://travis-ci.org/theislab/scvelo.svg?branch=master
+   :target: https://travis-ci.org/theislab/scvelo
+
 .. _velocyto: http://velocyto.org/
 .. _scanpy: https://github.com/theislab/scanpy
 

scvelo/datasets.py

@@ -1,7 +1,8 @@
 """Builtin Datasets.
 """
 
-from scanpy.api import read
+import scanpy.api as sc
+from scipy import sparse, stats
 
 
 def dentategyrus():
@@ -12,5 +13,13 @@ def dentategyrus():
     """
     filename = 'data/DentateGyrus/10X43_1.loom'
     url = 'http://pklab.med.harvard.edu/velocyto/DG1/10X43_1.loom'
-    adata = read(filename, backup_url=url, cleanup=True, sparse=True, cache=True)
+    adata = sc.read(filename, backup_url=url, cleanup=True, sparse=True, cache=True)
+    return adata
+
+
+def toy_data(n_obs, n_vars):
+    adata = sc.AnnData(sparse.random(n_obs, n_vars, data_rvs=stats.poisson(1).rvs, density=.6, format='csr'))
+    adata.layers['spliced'] = adata.X
+    adata.layers['unspliced'] = \
+       .5 * adata.X + .3 * sparse.random(n_obs, n_vars, density=.6, data_rvs=stats.poisson(1).rvs, format='csr')
     return adata

scvelo/preprocessing/__init__.py

@@ -46,4 +46,15 @@ def read_loom_layers(file_name, backup_url=None):
 
         adata = AnnData(X, obs=obs, var=var, layers=layers)
 
-    return adata
+    return adata
+
+
+def recipe_velocity(adata):
+    from scanpy.api.pp import \
+        filter_genes, filter_genes_dispersion, normalize_per_cell, pca, neighbors
+    filter_genes(adata, min_counts=10)
+    filter_genes_dispersion(adata)
+    normalize_per_cell(adata, layers='all')
+    pca(adata, n_comps=10)
+    neighbors(adata, use_rep='X_pca')
+    moments(adata)

tests/test_basic.py

@@ -1,40 +1,25 @@
+from scvelo.tools.utils import prod_sum_obs, prod_sum_var, norm
+import numpy as np
 import scvelo as scv
-import scanpy.api as sc
-from scvelo.tools.utils import *
-from scipy import sparse, stats
-
-
-def test_data(n_obs=1000, n_vars=100):
-    adata = sc.AnnData(sparse.random(n_obs, n_vars, data_rvs=stats.poisson(1).rvs, density=.6, format='csr'))
-    adata.layers['spliced'] = adata.X
-    adata.layers['unspliced'] = \
-       .5 * adata.X + .3 * sparse.random(n_obs, n_vars, density=.6, data_rvs=stats.poisson(1).rvs, format='csr')
-
-    sc.pp.filter_genes(adata, min_counts=10)
-    sc.pp.filter_genes_dispersion(adata)
-    sc.pp.normalize_per_cell(adata, layers='all')
-    sc.pp.pca(adata, n_comps=10)
-    sc.pp.neighbors(adata, use_rep='X_pca')
-    scv.pp.moments(adata)
-    return adata
 
 
 def test_einsum():
-    adata = test_data()
+    adata = scv.datasets.toy_data(n_obs=100, n_vars=100)
+    adata = scv.pp.recipe_velocity(adata)
     Ms, Mu = adata.layers['Ms'], adata.layers['Mu']
     assert np.allclose(prod_sum_obs(Ms, Mu), np.sum(Ms * Mu, 0))
     assert np.allclose(prod_sum_var(Ms, Mu), np.sum(Ms * Mu, 1))
     assert np.allclose(norm(Ms), np.linalg.norm(Ms, axis=1))
 
 
-# def test_velocity_graph():
-#     adata = test_data()
-#     scv.tl.velocity_graph(adata)
-#
-#     graph1 = adata.uns['velocity_graph'].copy()
-#
-#     scv.tl.velocity_graph(adata, n_jobs=2)
-#     graph2 = adata.uns['velocity_graph'].copy()
-#
-#     assert np.allclose((scv.tl.transition_matrix(adata) > 0).toarray(), (graph1 > 0).toarray())
-#     assert np.allclose(graph1.toarray(), graph2.toarray())
+def test_velocity_graph():
+    adata = scv.datasets.toy_data(n_obs=1000, n_vars=100)
+    scv.tl.velocity_graph(adata)
+
+    graph1 = adata.uns['velocity_graph'].copy()
+
+    scv.tl.velocity_graph(adata, n_jobs=2)
+    graph2 = adata.uns['velocity_graph'].copy()
+
+    assert np.allclose((scv.tl.transition_matrix(adata) > 0).toarray(), (graph1 > 0).toarray())
+    assert np.allclose(graph1.toarray(), graph2.toarray())