Make louvain optional for tests (scverse#2063)

* Make louvain dep optional for CI

* Make clustering tests relying on louvain optional

* Make louvain optional in test_neighbors_key_added.py

* Make logreg rank_genes_groups tests not require louvain

* Skip paul15 paga tests if louvain not installed

* Replace louvain with leiden in pbmc3k test

* Fix doc builds by updating mpl.colorbar.ColorBarBase to mpl.colorbar.ColorBar

.azure-pipelines.yml

@@ -43,7 +43,7 @@ jobs:
   - script: |
       python -m pip install --upgrade pip
       pip install pytest-cov wheel
-      pip install .[dev,test,louvain,leiden,magic,harmony,scrublet,scanorama,skmisc]
+      pip install .[dev,test,leiden,magic,harmony,scrublet,scanorama,skmisc]
     displayName: 'Install dependencies'
 
   - script: |

scanpy/plotting/_baseplot_class.py

@@ -507,7 +507,7 @@ def _plot_colorbar(self, color_legend_ax: Axes, normalize):
         cmap = pl.get_cmap(self.cmap)
         import matplotlib.colorbar
 
-        matplotlib.colorbar.ColorbarBase(
+        matplotlib.colorbar.Colorbar(
             color_legend_ax, orientation='horizontal', cmap=cmap, norm=normalize
         )
 

scanpy/plotting/_tools/paga.py

@@ -414,7 +414,7 @@ def paga(
     cax
         A matplotlib axes object for a potential colorbar.
     cb_kwds
-        Keyword arguments for :class:`~matplotlib.colorbar.ColorbarBase`,
+        Keyword arguments for :class:`~matplotlib.colorbar.Colorbar`,
         for instance, `ticks`.
     add_pos
         Add the positions to `adata.uns['paga']`.

scanpy/tests/notebooks/test_paga_paul15_subsampled.py

@@ -7,6 +7,7 @@
 
 import numpy as np
 from matplotlib.testing import setup
+import pytest
 
 setup()
 
@@ -40,6 +41,9 @@ def test_paga_paul15_subsampled(image_comparer, plt):
 
     sc.pl.draw_graph(adata, color='paul15_clusters', legend_loc='on data')
 
+    # TODO: skip if louvain isn't installed, needs major rework
+    pytest.importorskip("louvain")
+
     # Clustering and PAGA
     sc.tl.louvain(adata, resolution=1.0)
     sc.tl.paga(adata, groups='louvain')

scanpy/tests/notebooks/test_pbmc3k.py

@@ -105,23 +105,23 @@ def test_pbmc3k(image_comparer):
 
     # Clustering the graph
 
-    sc.tl.louvain(adata)
-    # sc.pl.umap(adata, color=['louvain', 'CST3', 'NKG7'], show=False)
+    sc.tl.leiden(adata)
+    # sc.pl.umap(adata, color=['leiden', 'CST3', 'NKG7'], show=False)
     # save_and_compare_images('umap_2')
-    sc.pl.scatter(adata, 'CST3', 'NKG7', color='louvain', show=False)
+    sc.pl.scatter(adata, 'CST3', 'NKG7', color='leiden', show=False)
     save_and_compare_images('scatter_3')
 
     # Finding marker genes
 
-    sc.tl.rank_genes_groups(adata, 'louvain')
+    sc.tl.rank_genes_groups(adata, 'leiden')
     sc.pl.rank_genes_groups(adata, n_genes=20, sharey=False, show=False)
     save_and_compare_images('rank_genes_groups_1')
 
-    sc.tl.rank_genes_groups(adata, 'louvain', method='logreg')
+    sc.tl.rank_genes_groups(adata, 'leiden', method='logreg')
     sc.pl.rank_genes_groups(adata, n_genes=20, sharey=False, show=False)
     save_and_compare_images('rank_genes_groups_2')
 
-    sc.tl.rank_genes_groups(adata, 'louvain', groups=['0'], reference='1')
+    sc.tl.rank_genes_groups(adata, 'leiden', groups=['0'], reference='1')
     sc.pl.rank_genes_groups(adata, groups='0', n_genes=20, show=False)
     save_and_compare_images('rank_genes_groups_3')
 
@@ -139,12 +139,12 @@ def test_pbmc3k(image_comparer):
         'Dendritic cells',
         'Megakaryocytes',
     ]
-    adata.rename_categories('louvain', new_cluster_names)
+    adata.rename_categories('leiden', new_cluster_names)
 
-    # sc.pl.umap(adata, color='louvain', legend_loc='on data', title='', frameon=False, show=False)
+    # sc.pl.umap(adata, color='leiden', legend_loc='on data', title='', frameon=False, show=False)
     # save_and_compare_images('umap_3')
 
     sc.pl.violin(
-        adata, ['CST3', 'NKG7', 'PPBP'], groupby='louvain', rotation=90, show=False
+        adata, ['CST3', 'NKG7', 'PPBP'], groupby='leiden', rotation=90, show=False
     )
     save_and_compare_images('violin_2')

scanpy/tests/test_clustering.py

@@ -19,6 +19,9 @@ def test_leiden_basic(adata_neighbors):
     ],
 )
 def test_clustering_subset(adata_neighbors, clustering, key):
+    if clustering == sc.tl.louvain:
+        pytest.importorskip("louvain")
+
     clustering(adata_neighbors, key_added=key)
 
     for c in adata_neighbors.obs[key].unique():
@@ -46,14 +49,16 @@ def test_clustering_subset(adata_neighbors, clustering, key):
 
 
 def test_louvain_basic(adata_neighbors):
+    pytest.importorskip("louvain")
+
     sc.tl.louvain(adata_neighbors)
     sc.tl.louvain(adata_neighbors, use_weights=True)
     sc.tl.louvain(adata_neighbors, use_weights=True, flavor="igraph")
     sc.tl.louvain(adata_neighbors, flavor="igraph")
 
 
 def test_partition_type(adata_neighbors):
-    import louvain
+    louvain = pytest.importorskip("louvain")
 
     sc.tl.louvain(adata_neighbors, partition_type=louvain.RBERVertexPartition)
     sc.tl.louvain(adata_neighbors, partition_type=louvain.SurpriseVertexPartition)

scanpy/tests/test_neighbors_key_added.py

@@ -52,12 +52,6 @@ def test_neighbors_key_obsp(adata, field):
     assert adata.uns['leiden']['params'] == adata1.uns['leiden']['params']
     assert np.all(adata.obs['leiden'] == adata1.obs['leiden'])
 
-    sc.tl.louvain(adata, random_state=0)
-    sc.tl.louvain(adata1, random_state=0, **arg)
-
-    assert adata.uns['louvain']['params'] == adata1.uns['louvain']['params']
-    assert np.all(adata.obs['louvain'] == adata1.obs['louvain'])
-
     # no obsp in umap, paga
     if field == 'neighbors_key':
         sc.tl.umap(adata, random_state=0)
@@ -77,3 +71,23 @@ def test_neighbors_key_obsp(adata, field):
             adata.uns['paga']['connectivities_tree'].toarray(),
             adata1.uns['paga']['connectivities_tree'].toarray(),
         )
+
+
+@pytest.mark.parametrize('field', ['neighbors_key', 'obsp'])
+def test_neighbors_key_obsp_louvain(adata, field):
+    pytest.importorskip("louvain")
+    adata1 = adata.copy()
+
+    sc.pp.neighbors(adata, n_neighbors=n_neighbors, random_state=0)
+    sc.pp.neighbors(adata1, n_neighbors=n_neighbors, random_state=0, key_added=key)
+
+    if field == 'neighbors_key':
+        arg = {field: key}
+    else:
+        arg = {field: adata1.uns[key]['connectivities_key']}
+
+    sc.tl.louvain(adata, random_state=0)
+    sc.tl.louvain(adata1, random_state=0, **arg)
+
+    assert adata.uns['louvain']['params'] == adata1.uns['louvain']['params']
+    assert np.all(adata.obs['louvain'] == adata1.obs['louvain'])

scanpy/tests/test_rank_genes_groups_logreg.py

@@ -1,44 +1,43 @@
+import pytest
+
 import numpy as np
 import scanpy as sc
 
 
-def test_rank_genes_groups_with_renamed_categories():
+@pytest.mark.parametrize(
+    "method",
+    ["t-test", "logreg"],
+)
+def test_rank_genes_groups_with_renamed_categories(method):
     adata = sc.datasets.blobs(n_variables=4, n_centers=3, n_observations=200)
     assert np.allclose(adata.X[1], [9.214668, -2.6487126, 4.2020774, 0.51076424])
 
-    sc.tl.pca(adata)
-    sc.pp.neighbors(adata)
-
-    for method in ['logreg', 't-test']:
-        sc.tl.louvain(adata)
+    # for method in ['logreg', 't-test']:
 
-        sc.tl.rank_genes_groups(adata, 'louvain', method=method)
-        assert adata.uns['rank_genes_groups']['names'].dtype.names == ('0', '1', '2')
-        assert adata.uns['rank_genes_groups']['names'][0].tolist() == ('3', '1', '0')
+    sc.tl.rank_genes_groups(adata, 'blobs', method=method)
+    assert adata.uns['rank_genes_groups']['names'].dtype.names == ('0', '1', '2')
+    assert adata.uns['rank_genes_groups']['names'][0].tolist() == ('1', '3', '0')
 
-        adata.rename_categories('louvain', ['Zero', 'One', 'Two'])
-        assert adata.uns['rank_genes_groups']['names'][0].tolist() == ('3', '1', '0')
+    adata.rename_categories('blobs', ['Zero', 'One', 'Two'])
+    assert adata.uns['rank_genes_groups']['names'][0].tolist() == ('1', '3', '0')
 
-        sc.tl.rank_genes_groups(adata, 'louvain', method=method)
-        assert adata.uns['rank_genes_groups']['names'][0].tolist() == ('3', '1', '0')
+    sc.tl.rank_genes_groups(adata, 'blobs', method=method)
+    assert adata.uns['rank_genes_groups']['names'][0].tolist() == ('1', '3', '0')
+    assert adata.uns['rank_genes_groups']['names'].dtype.names == ('Zero', 'One', 'Two')
 
 
 def test_rank_genes_groups_with_renamed_categories_use_rep():
     adata = sc.datasets.blobs(n_variables=4, n_centers=3, n_observations=200)
     assert np.allclose(adata.X[1], [9.214668, -2.6487126, 4.2020774, 0.51076424])
 
-    sc.tl.pca(adata)
-    sc.pp.neighbors(adata)
-
     adata.layers["to_test"] = adata.X.copy()
     adata.X = adata.X[::-1, :]
 
-    sc.tl.louvain(adata)
     sc.tl.rank_genes_groups(
-        adata, 'louvain', method='logreg', layer="to_test", use_raw=False
+        adata, 'blobs', method='logreg', layer="to_test", use_raw=False
     )
     assert adata.uns['rank_genes_groups']['names'].dtype.names == ('0', '1', '2')
-    assert adata.uns['rank_genes_groups']['names'][0].tolist() == ('3', '1', '0')
+    assert adata.uns['rank_genes_groups']['names'][0].tolist() == ('1', '3', '0')
 
-    sc.tl.rank_genes_groups(adata, 'louvain', method="logreg")
+    sc.tl.rank_genes_groups(adata, 'blobs', method="logreg")
     assert not adata.uns['rank_genes_groups']['names'][0].tolist() == ('3', '1', '0')