Add tskit zarr_format test

Add plink zarr_format test

Add vcf zarr_format test

Author: tomwhite

tests/test_plink.py

@@ -443,6 +443,15 @@ def test_by_validating(
     validate(path, out)
 
 
+@pytest.mark.parametrize("zarr_format", [2, 3])
+def test_zarr_format(tmp_path, zarr_format):
+    path = "tests/data/plink/plink_sim_10s_100v_10pmiss"
+    out = tmp_path / "example.zarr"
+    root = plink.convert(path, out, zarr_format=zarr_format)
+    assert root.metadata.zarr_format == zarr_format
+    validate(path, out)
+
+
 class TestMultipleContigs:
     """Test handling of multiple contigs in PLINK files."""
 

tests/test_tskit.py

@@ -575,3 +575,13 @@ def test_workers(tmp_path, worker_processes):
     root = tsk.convert(ts, worker_processes=worker_processes)
     ds = load_dataset(root)
     assert_ts_ds_equal(ts, ds)
+
+
+@pytest.mark.parametrize("zarr_format", [2, 3])
+def test_zarr_format(zarr_format):
+    ts = msprime.sim_ancestry(10, sequence_length=1000, random_seed=42)
+    ts = add_mutations(ts)
+    root = tsk.convert(ts, zarr_format=zarr_format)
+    assert root.metadata.zarr_format == zarr_format
+    ds = load_dataset(root)
+    assert_ts_ds_equal(ts, ds)

tests/test_vcf_examples.py

@@ -394,6 +394,16 @@ def test_worker_processes(self, ds, worker_processes):
         ds2 = load_dataset(root)
         assert_dataset_equal(ds, ds2, drop_vars=["region_index"])
 
+    @pytest.mark.parametrize("zarr_format", [2, 3])
+    def test_zarr_format(self, ds, zarr_format):
+        root = vcf_mod.convert(
+            [self.data_path],
+            zarr_format=zarr_format,
+        )
+        assert root.metadata.zarr_format == zarr_format
+        ds2 = load_dataset(root)
+        assert_dataset_equal(ds, ds2, drop_vars=["region_index"])
+
     def test_inspect(self, tmp_path):
         # TODO pretty weak test, we should be doing this better somewhere else
         out = tmp_path / "example.vcf.zarr"