Bump up test coverage

Author: jeromekelleher

bio2zarr/tskit.py

@@ -45,9 +45,7 @@ def __init__(
         if self._num_samples < 1:
             raise ValueError("individuals_nodes must have at least one sample")
         self.max_ploidy = individuals_nodes.shape[1]
-        if sample_ids is None:
-            sample_ids = [f"tsk_{j}" for j in range(self._num_samples)]
-        elif len(sample_ids) != self._num_samples:
+        if len(sample_ids) != self._num_samples:
             raise ValueError(
                 f"Length of sample_ids ({len(sample_ids)}) does not match "
                 f"number of samples ({self._num_samples})"

tests/test_tskit.py

@@ -29,6 +29,12 @@ def test_missing_dependency():
         )
 
 
+def tskit_model_mapping(ind_nodes, ind_names=None):
+    if ind_names is None:
+        ind_names = ["tsk{j}" for j in range(len(ind_nodes))]
+    return tskit.VcfModelMapping(ind_nodes, ind_names)
+
+
 def add_mutations(ts):
     # Add some mutation to the tree sequence. This guarantees that
     # we have variation at all sites > 0.
@@ -311,6 +317,23 @@ def test_iter_field(self, fx_simple_ts):
         with pytest.raises(ValueError, match="Unknown field"):
             list(format_obj.iter_field("unknown_field", None, 0, 3))
 
+    def test_zero_samples(self, fx_simple_ts):
+        model_mapping = tskit_model_mapping(np.array([]))
+        with pytest.raises(ValueError, match="at least one sample"):
+            tsk.TskitFormat(fx_simple_ts, model_mapping=model_mapping)
+
+    def test_no_valid_samples(self, fx_simple_ts):
+        model_mapping = fx_simple_ts.map_to_vcf_model()
+        model_mapping.individuals_nodes[:] = -1
+        with pytest.raises(ValueError, match="at least one valid sample"):
+            tsk.TskitFormat(fx_simple_ts, model_mapping=model_mapping)
+
+    def test_model_size_mismatch(self, fx_simple_ts):
+        model_mapping = fx_simple_ts.map_to_vcf_model()
+        model_mapping.individuals_name = ["x"]
+        with pytest.raises(ValueError, match="match number of samples"):
+            tsk.TskitFormat(fx_simple_ts, model_mapping=model_mapping)
+
     @pytest.mark.parametrize(
         ("ind_nodes", "expected_gts"),
         [
@@ -347,10 +370,7 @@ def test_iter_field(self, fx_simple_ts):
         ],
     )
     def test_iter_alleles_and_genotypes(self, fx_simple_ts, ind_nodes, expected_gts):
-        model_mapping = tskit.VcfModelMapping(
-            ind_nodes, ["tsk{j}" for j in range(len(ind_nodes))]
-        )
-
+        model_mapping = tskit_model_mapping(ind_nodes)
         format_obj = tsk.TskitFormat(fx_simple_ts, model_mapping=model_mapping)
 
         shape = (2, 2)  # (num_samples, max_ploidy)
@@ -375,9 +395,7 @@ def test_iter_alleles_and_genotypes(self, fx_simple_ts, ind_nodes, expected_gts)
     def test_iter_alleles_and_genotypes_missing_node(self, fx_ts_2_diploids):
         # Test with node ID that doesn't exist in tree sequence (out of range)
         ind_nodes = np.array([[10, 11], [12, 13]], dtype=np.int32)
-        model_mapping = tskit.VcfModelMapping(
-            ind_nodes, ["tsk{j}" for j in range(len(ind_nodes))]
-        )
+        model_mapping = tskit_model_mapping(ind_nodes)
         format_obj = tsk.TskitFormat(fx_ts_2_diploids, model_mapping=model_mapping)
         shape = (2, 2)
         with pytest.raises(
@@ -387,9 +405,7 @@ def test_iter_alleles_and_genotypes_missing_node(self, fx_ts_2_diploids):
 
     def test_isolated_as_missing(self, fx_ts_isolated_samples):
         ind_nodes = np.array([[0], [1], [3]])
-        model_mapping = tskit.VcfModelMapping(
-            ind_nodes, ["tsk{j}" for j in range(len(ind_nodes))]
-        )
+        model_mapping = tskit_model_mapping(ind_nodes)
 
         format_obj_default = tsk.TskitFormat(
             fx_ts_isolated_samples,