Merge pull request #18 from hillsbury/issues/15

added assertion error for missing replicates

cytominer_eval/operations/percent_strong.py

@@ -29,6 +29,7 @@ def percent_strong(
     A metric describing the proportion of replicates that correlate above the given
     quantile of non-replicate correlation distribution
     """
+
     assert 0 < quantile and 1 >= quantile, "quantile must be between 0 and 1"
 
     similarity_melted_df = assign_replicates(
@@ -38,13 +39,18 @@ def percent_strong(
     # Check to make sure that the melted dataframe is upper triangle
     assert_melt(similarity_melted_df, eval_metric="percent_strong")
 
+    # check that there are group_replicates (non-unique rows)
+    replicate_df = similarity_melted_df.query("group_replicate")
+    denom = replicate_df.shape[0]
+
+    assert denom != 0, "no replicate groups identified in {rep} columns!".format(
+        rep=replicate_groups
+    )
+
     non_replicate_quantile = similarity_melted_df.query(
         "not group_replicate"
     ).similarity_metric.quantile(quantile)
 
-    replicate_df = similarity_melted_df.query("group_replicate")
-    denom = replicate_df.shape[0]
-
     percent_strong = (
         replicate_df.similarity_metric > non_replicate_quantile
     ).sum() / denom

cytominer_eval/tests/test_operations/test_percent_strong.py

@@ -53,3 +53,16 @@ def test_percent_strong():
     expected_result = 0.3074
 
     assert np.round(output, 4) == expected_result
+
+
+def test_percent_strong_uniquerows():
+    with pytest.raises(AssertionError) as err:
+        replicate_groups = ["Metadata_pert_well"]
+        output = percent_strong(
+            similarity_melted_df=similarity_melted_df,
+            replicate_groups=replicate_groups,
+            quantile=0.95,
+        )
+    assert "no replicate groups identified in {rep} columns!".format(
+        rep=replicate_groups
+    ) in str(err.value)