Fix VCF tests

Author: benjeffery

python/tests/test_vcf.py

@@ -357,21 +357,18 @@ def test_individuals_no_nodes_default_args(self):
         with warnings.catch_warnings(record=True) as w:
             with pytest.raises(ValueError, match="No samples in resulting VCF model"):
                 ts2.as_vcf(allow_position_zero=True)
-            assert len(w) == 2
+            assert len(w) == 1
             assert "At least one sample node does not have an individual ID" in str(
                 w[0].message
             )
-            assert "Individual 0 has no nodes associated with it." in str(w[1].message)
 
     def test_individuals_no_nodes_as_argument(self):
         ts1 = msprime.simulate(10, mutation_rate=0.1, random_seed=2)
         tables = ts1.dump_tables()
         tables.individuals.add_row()
         ts2 = tables.tree_sequence()
         with warnings.catch_warnings(record=True) as w:
-            with pytest.raises(
-                ValueError, match="Individual 0 has no nodes associated with it."
-            ):
+            with pytest.raises(ValueError, match="No samples in resulting VCF model"):
                 ts2.as_vcf(individuals=[0])
             assert len(w) == 1
             assert "At least one sample node does not have an individual ID" in str(
@@ -395,7 +392,7 @@ def test_ploidy_with_no_node_individuals(self):
     def test_empty_individuals(self):
         ts = msprime.sim_ancestry(3, random_seed=2)
         ts = tsutil.insert_branch_sites(ts)
-        with pytest.raises(ValueError, match="No samples in resulting VCF model"):
+        with pytest.raises(ValueError, match="No individuals specified"):
             ts.as_vcf(individuals=[])
 
     def test_duplicate_individuals(self):
@@ -404,28 +401,6 @@ def test_duplicate_individuals(self):
         with pytest.raises(tskit.LibraryError, match="TSK_ERR_DUPLICATE_SAMPLE"):
             ts.as_vcf(individuals=[0, 0], allow_position_zero=True)
 
-    def test_mixed_sample_non_sample_individuals(self):
-        ts = msprime.sim_ancestry(3, random_seed=2)
-        tables = ts.dump_tables()
-        tables.individuals.add_row()
-        # Add a reference to an individual from a non-sample
-        individual = tables.nodes.individual
-        individual[-1] = 0
-        tables.nodes.individual = individual
-        ts = tables.tree_sequence()
-        ts = tsutil.insert_branch_sites(ts)
-        with warnings.catch_warnings(record=True) as w:
-            ts.map_to_vcf_model()
-            assert len(w) == 2
-            assert (
-                "Individual 0 has both sample and non-sample nodes associated with it."
-                in str(w[0].message)
-            )
-            assert "Individual 3 has no nodes associated with it." in str(w[1].message)
-        with warnings.catch_warnings(record=True) as w:
-            assert len(ts.as_vcf(individuals=[1, 2], allow_position_zero=True)) > 0
-            assert len(w) == 0
-
     def test_samples_with_and_without_individuals(self):
         ts = tskit.Tree.generate_balanced(3).tree_sequence
         tables = ts.dump_tables()

python/tskit/trees.py

@@ -10604,6 +10604,8 @@ def map_to_vcf_model(
         else:
             if individuals is None:
                 individuals = np.arange(self.num_individuals, dtype=np.int32)
+            if len(individuals) == 0:
+                raise ValueError("No individuals specified")
             if min(individuals) < 0 or max(individuals) >= self.num_individuals:
                 raise ValueError("Invalid individual ID")
 

python/tskit/vcf.py

@@ -71,18 +71,24 @@ def __init__(
         vcf_model = tree_sequence.map_to_vcf_model(
             individuals=individuals, ploidy=ploidy, individual_names=individual_names
         )
-        self.individual_names = vcf_model.individuals_name
+        # Remove individuals with zero ploidy as these cannot be
+        # represented in VCF.
+        individuals_nodes = vcf_model.individuals_nodes
+        to_keep = (individuals_nodes != -1).any(axis=1)
+        individuals_nodes = individuals_nodes[to_keep]
+        self.individual_names = vcf_model.individuals_name[to_keep]
+
         self.individual_ploidies = [
             len(nodes[nodes >= 0]) for nodes in vcf_model.individuals_nodes
         ]
         self.num_individuals = len(self.individual_names)
 
-        if len(vcf_model.individuals_nodes) == 0:
+        if len(individuals_nodes) == 0:
             raise ValueError("No samples in resulting VCF model")
 
         # Flatten the array of node IDs, filtering out the -1 padding values
         self.samples = []
-        for row in vcf_model.individuals_nodes:
+        for row in individuals_nodes:
             for node_id in row:
                 if node_id != -1:
                     self.samples.append(node_id)