Fix cnv_discordant_read_calls: raise ValueError when no data found

malariagen_data/anoph/cnv_data.py

@@ -610,13 +610,19 @@ def cnv_discordant_read_calls(
 
                 ly.append(y)
 
-                if len(ly) == 0:
-                    # Bail out, no data for given sample sets and analysis.
-                    raise ValueError("No data found for requested sample sets.")
+            # Check after processing all sample sets for a given contig.
+            if not ly:
+                # Bail out, no data for given sample sets and analysis.
+                raise ValueError("No data found for requested sample sets.")
 
             x = simple_xarray_concat(ly, dim=DIM_SAMPLE)
             lx.append(x)
 
+        # Optionally, check if no contigs yielded data.
+        if not lx:
+            raise ValueError("No data found for requested sample sets across all contigs.")
+
+
         ds = simple_xarray_concat(lx, dim=DIM_VARIANT)
 
         debug("handle sample query")

malariagen_data/util.py

@@ -488,7 +488,6 @@ def init_filesystem(url, **kwargs):
 
     # Process the URL using fsspec.
     fs, path = url_to_fs(url, **storage_options)
-
     # Path compatibility, fsspec/gcsfs behaviour varies between versions.
     while path.endswith("/"):
         path = path[:-1]

tests/anoph/test_cnv_data.py

@@ -181,6 +181,21 @@ def test_open_cnv_discordant_read_calls(fixture, api: AnophelesCnvData):
         root = api.open_cnv_discordant_read_calls(sample_set="foobar")
 
 
+def fake_none_dataset(*args, **kwargs):
+    return None
+
+def test_cnv_discordant_read_calls_no_data(monkeypatch):
+    # Create an instance of AnophelesCnvData; assuming you have a fixture or a way to instantiate.
+    api = AnophelesCnvData(...)
+
+    # Monkey-patch the dataset method to always return None.
+    monkeypatch.setattr(api, "_cnv_discordant_read_calls_dataset", fake_none_dataset)
+
+    # Expect a ValueError since no dataset will be found.
+    with pytest.raises(ValueError, match="No data found for requested sample sets"):
+        # Pass some valid contig string (e.g., "2RL") and sample_sets
+        api.cnv_discordant_read_calls(contig="2RL", sample_sets=["some_sample_set"])
+
 def test_cnv_hmm__sample_query(ag3_sim_fixture, ag3_sim_api: AnophelesCnvData):
     api = ag3_sim_api
     fixture = ag3_sim_fixture