Merge pull request #12 from YosefLab/dist-patch

fixed distance with dataframes

src/pycea/tl/distance.py

@@ -109,6 +109,8 @@ def distance(
     connect_key = _format_keys(connect_key, "connectivities")
     single_obs = False
     X = get_keyed_obsm_data(tdata, key)
+    if isinstance(X, pd.DataFrame):
+        X = X.values
     if update:
         _check_previous_params(
             tdata, {"metric": metric, "metric_kwds": metric_kwds}, key_added, ["neighbors", "distances"]

tests/test_distance.py

@@ -11,7 +11,10 @@
 def tdata():
     tdata = td.TreeData(
         obs=pd.DataFrame({"group": ["1", "1", "2"]}, index=["A", "B", "C"]),
-        obsm={"spatial": np.array([[0, 0], [1, 1], [2, 2]]), "characters": np.array([[0, 0], [1, 1], [0, 1]])},
+        obsm={
+            "spatial": np.array([[0, 0], [1, 1], [2, 2]]),
+            "characters": pd.DataFrame(np.array([[0, 0], [1, 1], [0, 1]]), index=["A", "B", "C"], columns=["c1", "c2"]),
+        },
         obsp={"connectivities": sp.sparse.csr_matrix(([1, 1], ([0, 0], [1, 2])), shape=(3, 3))},
     )
     yield tdata
@@ -63,6 +66,10 @@ def test_sampled_distance(tdata):
     assert tdata.obsp["spatial_distances"].data.tolist() == [2, 0]
     assert tdata.obsp["spatial_connectivities"].shape == (3, 3)
     assert len(tdata.obsp["spatial_connectivities"].data) == 2
+    distance(tdata, "characters", sample_n=2, metric="cityblock", random_state=3)
+    assert tdata.obsp["characters_distances"].shape == (3, 3)
+    assert len(tdata.obsp["characters_distances"].data) == 2
+    assert tdata.obsp["characters_distances"].data.tolist() == [1, 1]
 
 
 def test_connected_distance(tdata):