comment cleanup

parashardhapola · Oct 29, 2024 · 9fccff9 · 9fccff9
1 parent c90134c
commit 9fccff9
Show file tree

Hide file tree

Showing 2 changed files with 16 additions and 7 deletions.
diff --git a/scarf/metrics.py b/scarf/metrics.py
@@ -19,7 +19,6 @@
 
 # LISI - The Local Inverse Simpson Index
 def compute_lisi(
-    # X: np.array,
     distances: zarrArrayType,
     indices: zarrArrayType,
     metadata: pd.DataFrame,
@@ -56,7 +55,6 @@ def compute_lisi(
     # Don't count yourself
     indices = indices[:, 1:]
     distances = distances[:, 1:]
-    # Save the result
     lisi_df = np.zeros((n_cells, n_labels))
     for i, label in enumerate(label_colnames):
         logger.info(f"Computing LISI for {label}")
@@ -66,7 +64,6 @@ def compute_lisi(
             distances.T, indices.T, labels, n_categories, perplexity
         )
         lisi_df[:, i] = 1 / simpson
-    # lisi_df = lisi_df.flatten()
     return lisi_df
 
 

diff --git a/scarf/tests/test_metrics.py b/scarf/tests/test_metrics.py
@@ -2,9 +2,6 @@
 
 
 def test_metric_lisi(datastore, make_graph):
-    # datastore.auto_filter_cells(show_qc_plots=False)
-    # datastore.mark_hvgs(top_n=100, show_plot=False)
-    # datastore.make_graph(feat_key="hvgs")
     lables = np.random.randint(0, 2, datastore.cells.N)
     datastore.cells.insert(
         column_name="samples_id",
@@ -21,7 +18,7 @@ def test_metric_silhouette(datastore, make_graph, leiden_clustering):
     _ = datastore.metric_silhouette()
 
 
-def test_metric_integration(datastore, make_graph, leiden_clustering):
+def test_metric_integration_ari(datastore, make_graph, leiden_clustering):
     lables1 = np.random.randint(0, 2, datastore.cells.N)
     lables2 = np.random.randint(0, 2, datastore.cells.N)
     datastore.cells.insert(
@@ -35,3 +32,18 @@ def test_metric_integration(datastore, make_graph, leiden_clustering):
         overwrite=True,
     )
     _ = datastore.metric_integration(batch_labels=["lables1", "lables2"], metric="ari")
+
+def test_metric_integration_nmi(datastore, make_graph, leiden_clustering):
+    lables1 = np.random.randint(0, 2, datastore.cells.N)
+    lables2 = np.random.randint(0, 2, datastore.cells.N)
+    datastore.cells.insert(
+        column_name="lables1",
+        values=lables1,
+        overwrite=True,
+    )
+    datastore.cells.insert(
+        column_name="lables2",
+        values=lables2,
+        overwrite=True,
+    )
+    _ = datastore.metric_integration(batch_labels=["lables1", "lables2"], metric="nmi")