Merge pull request #3 from pinellolab/expand_test

Expand testing

.github/workflows/CI.yml

@@ -25,7 +25,7 @@ jobs:
         conda config --add channels bioconda 
         conda config --add channels conda-forge
         # conda env update --file environment.yml --name base
-        conda install simba_pbg=1.1
+        conda install simba
         python -m pip install --upgrade pip
         pip install -r requirements.txt
         pip install -e .

simba/plotting/_plot.py

@@ -1182,6 +1182,9 @@ def discretize(adata,
 
     assert 'disc' in adata.uns_keys(), \
         "please run `si.tl.discretize()` first"
+    if kde is not None:
+        warnings.warn("kde is no longer supported as of v1.1",
+                      DeprecationWarning)
 
     hist_edges = adata.uns['disc']['hist_edges']
     hist_count = adata.uns['disc']['hist_count']

tests/test_pbg_training.py

@@ -21,6 +21,8 @@ def test_pbg_training_rna(adata_CG, tmp_path):
                     dirname='graph0')
     si.tl.pbg_train(auto_wd=True,
                     output='model')
+    si.pl.pbg_metrics(fig_ncol=1,
+                      save_fig=True)
 
 
 def test_pbg_training_atac(adata_CP, tmp_path):
@@ -30,3 +32,5 @@ def test_pbg_training_atac(adata_CP, tmp_path):
                     dirname='graph0')
     si.tl.pbg_train(auto_wd=True,
                     output='model')
+    si.pl.pbg_metrics(fig_ncol=1,
+                      save_fig=True)

tests/test_post_training.py

@@ -11,12 +11,54 @@ def dict_adata():
         path_entity_alias='tests/data/pbg_training')
 
 
-def test_embeddding_rna(dict_adata):
+def test_embeddding_rna(dict_adata, tmp_path):
+    si.settings.set_workdir(tmp_path / "simba_rna")
     adata_C = dict_adata['C']
     adata_G = dict_adata['G']
     adata_all_CG = si.tl.embed(
         adata_ref=adata_C,
         list_adata_query=[adata_G])
+    # add annotations of cells and genes
+    adata_all_CG.obs['entity_anno'] = ""
+    adata_all_CG.obs.loc[adata_C.obs_names, 'entity_anno'] = 'cell'
+    adata_all_CG.obs.loc[adata_G.obs_names, 'entity_anno'] = 'gene'
+
     si.tl.umap(adata_all_CG,
                n_neighbors=15,
                n_components=2)
+    adata_cmp = si.tl.compare_entities(
+        adata_ref=adata_C,
+        adata_query=adata_G)
+    si.pl.entity_metrics(adata_cmp,
+                         x='max',
+                         y='gini',
+                         show_contour=False,
+                         texts=adata_G.obs_names[:2],
+                         show_texts=True,
+                         show_cutoff=True,
+                         size=5,
+                         text_expand=(1.3, 1.5),
+                         cutoff_x=1.,
+                         cutoff_y=0.3,
+                         save_fig=True)
+    si.pl.entity_barcode(adata_cmp,
+                         layer='softmax',
+                         entities=list(adata_G.obs_names[:2]),
+                         show_cutoff=True,
+                         cutoff=0.001,
+                         fig_size=(5, 2.5),
+                         save_fig=True)
+    query_result = si.tl.query(adata_all_CG,
+                               entity=list(adata_C.obs_names[:2]),
+                               obsm='X_umap',
+                               use_radius=False,
+                               k=50,
+                               anno_filter='entity_anno',
+                               filters=['gene'])
+    print(query_result.head())
+    si.pl.query(adata_all_CG,
+                show_texts=False,
+                color=['entity_anno'],
+                alpha=0.9,
+                alpha_bg=0.1,
+                save_fig=True)

tests/test_preprocessing.py

@@ -12,18 +12,85 @@ def adata_CP():
     return si.read_h5ad("tests/data/10xpbmc_atac_subset.h5ad")
 
 
-def test_rna(adata_CG):
+def test_rna(adata_CG, tmp_path):
+    si.settings.set_workdir(tmp_path / "simba_rna")
+    si.settings.set_figure_params(dpi=80,
+                                  style='white',
+                                  fig_size=[5, 5],
+                                  rc={'image.cmap': 'viridis'})
     si.pp.filter_genes(adata_CG, min_n_cells=3)
     si.pp.cal_qc_rna(adata_CG)
+    si.pl.violin(adata_CG,
+                 list_obs=['n_counts', 'n_genes', 'pct_mt'],
+                 save_fig=True,
+                 fig_name='plot_violin.png')
+    si.pp.filter_cells_rna(adata_CG, min_n_genes=2)
     si.pp.normalize(adata_CG, method='lib_size')
     si.pp.log_transform(adata_CG)
     si.pp.select_variable_genes(adata_CG, n_top_genes=2000)
+    si.pl.variable_genes(adata_CG,
+                         show_texts=True,
+                         save_fig=True,
+                         fig_name='plot_variable_genes.png')
     si.tl.discretize(adata_CG, n_bins=5)
+    si.pl.discretize(adata_CG,
+                     save_fig=True,
+                     fig_name='plot_discretize.png')
 
 
-def test_atac(adata_CP):
-    si.pp.filter_peaks(adata_CP, min_n_cells=3)
+def test_atac(adata_CP, tmp_path):
+    si.settings.set_workdir(tmp_path / "simba_atac")
+    si.pp.filter_peaks(adata_CP, min_n_cells=5)
     si.pp.cal_qc_atac(adata_CP)
+    si.pl.hist(adata_CP,
+               list_obs=['n_counts', 'n_peaks', 'pct_peaks'],
+               log=True,
+               list_var=['n_cells'],
+               fig_size=(3, 3),
+               save_fig=True,
+               fig_name='plot_histogram.png')
+    si.pp.filter_cells_atac(adata_CP, min_n_peaks=5)
     si.pp.pca(adata_CP, n_components=30)
+    si.pl.pca_variance_ratio(adata_CP,
+                             show_cutoff=True,
+                             save_fig=True,
+                             fig_name='plot_variance_ratio.png')
     si.pp.select_pcs(adata_CP, n_pcs=10)
     si.pp.select_pcs_features(adata_CP)
+    si.pl.pcs_features(adata_CP,
+                       fig_ncol=5,
+                       save_fig=True,
+                       fig_name='plot_pcs_features.png')
+
+
+def test_genescores(adata_CP):
+    si.pp.filter_peaks(adata_CP, min_n_cells=5)
+    si.pp.cal_qc_atac(adata_CP)
+    si.pp.filter_cells_atac(adata_CP, min_n_peaks=5)
+    si.pp.pca(adata_CP, n_components=30)
+    si.pp.select_pcs(adata_CP, n_pcs=10)
+    si.pp.select_pcs_features(adata_CP)
+
+    adata_CG_atac = si.tl.gene_scores(adata_CP,
+                                      genome='hg19',
+                                      use_gene_weigt=True,
+                                      use_top_pcs=True)
+    print(adata_CG_atac)
+
+
+def test_integration(adata_CG):
+    si.pp.filter_genes(adata_CG, min_n_cells=3)
+    si.pp.cal_qc_rna(adata_CG)
+    si.pp.filter_cells_rna(adata_CG, min_n_genes=2)
+    si.pp.normalize(adata_CG, method='lib_size')
+    si.pp.log_transform(adata_CG)
+    si.pp.select_variable_genes(adata_CG, n_top_genes=2000)
+    adata_C1C2 = si.tl.infer_edges(
+        adata_CG, adata_CG, n_components=20, k=20)
+    si.pl.node_similarity(adata_C1C2,
+                          cutoff=0.5,
+                          save_fig=True)
+    si.pl.svd_nodes(adata_C1C2,
+                    cutoff=0.5,
+                    save_fig=True)
+    si.tl.trim_edges(adata_C1C2, cutoff=0.5)