Merge branch 'master' into GH953-synonymous-start-stop

Author: ZiadXI

malariagen_data/anoph/cnv_data.py

@@ -642,9 +642,13 @@ 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.")
+            if len(ly) == 0:
+                # Bail out, no data for given sample sets and contig.
+                raise ValueError(
+                    f"No CNV discordant read calls data found for contig {c!r} "
+                    f"in the requested sample sets. This could be because the "
+                    f"sample sets do not have discordant read calls data available."
+                )
 
             x = _simple_xarray_concat(ly, dim=DIM_SAMPLE)
             lx.append(x)

malariagen_data/anoph/fst.py

@@ -539,11 +539,10 @@ def plot_pairwise_average_fst(
             if annotation == "standard error":
                 fig_df.loc[cohort1, cohort2] = se
             elif annotation == "Z score":
-                try:
-                    zs = fst / se
-                    fig_df.loc[cohort1, cohort2] = zs
-                except ZeroDivisionError:
+                if se == 0:
                     fig_df.loc[cohort1, cohort2] = np.nan
+                else:
+                    fig_df.loc[cohort1, cohort2] = fst / se
             else:
                 fig_df.loc[cohort1, cohort2] = fst
 

malariagen_data/anoph/g123.py

@@ -170,7 +170,7 @@ def g123_gwss(
     ) -> Tuple[np.ndarray, np.ndarray]:
         # Change this name if you ever change the behaviour of this function, to
         # invalidate any previously cached data.
-        name = "g123_gwss_v1"
+        name = "g123_gwss_v2"
 
         valid_sites = self.phasing_analysis_ids + ("all", "segregating")
         if sites not in valid_sites:
@@ -181,7 +181,7 @@ def g123_gwss(
         params = dict(
             contig=contig,
             sites=sites,
-            site_mask=site_mask,
+            site_mask=self._prep_optional_site_mask_param(site_mask=site_mask),
             window_size=window_size,
             sample_sets=self._prep_sample_sets_param(sample_sets=sample_sets),
             # N.B., do not be tempted to convert this sample query into integer

malariagen_data/mjn.py

@@ -281,12 +281,6 @@ def _mjn_graph_edges(
                 # add edge from final intermediate node to node j
                 source = f"anon_{i}_{j}_{sep-2}"
                 target = j
-                graph_node = {
-                    "id": source,
-                    "count": 0,
-                    "width": anon_width,
-                }
-                graph_nodes.append(graph_node)
                 graph_edge = {
                     "id": f"edge_{i}_{j}_{sep-1}",
                     "source": source,

tests/anoph/test_cnv_data.py

@@ -626,7 +626,10 @@ def test_cnv_discordant_read_calls(fixture, api: AnophelesCnvData):
             assert isinstance(d2, xr.DataArray)
 
     # Check with a contig that should not exist
-    with pytest.raises(ValueError):
+    with pytest.raises(
+        ValueError,
+        match="No CNV discordant read calls data found|no CNVs available for contig",
+    ):
         api.cnv_discordant_read_calls(
             contig="foobar", sample_sets=random.choice(all_sample_sets)
         )

tests/anoph/test_g123.py

@@ -240,8 +240,7 @@ def test_g123_calibration(fixture, api: AnophelesG123Analysis):
 
     # Set up test parameters.
     all_sample_sets = api.sample_sets()["sample_set"].to_list()
-    window_sizes = np.random.randint(100, 500, size=random.randint(2, 5)).tolist()
-    window_sizes = sorted([int(x) for x in window_sizes])
+    window_sizes = sorted(random.sample(range(100, 500), k=random.randint(2, 5)))
     g123_params = dict(
         contig=random.choice(api.contigs),
         sites=random.choice(api.phasing_analysis_ids),