Switch tskit haploid phasing back to True

Author: jeromekelleher

bio2zarr/tskit.py

@@ -41,12 +41,6 @@ def __init__(
         individuals_nodes = model_mapping.individuals_nodes
         sample_ids = model_mapping.individuals_name
 
-        self.is_phased = True
-        if individuals_nodes.shape[1] == 1:
-            # For simplicity we defined haploids as unphased to do the same thing as the
-            # VCF conversion code. We should just omit the array for haploids anyway.
-            self.is_phased = False
-
         self._num_samples = individuals_nodes.shape[0]
         if self._num_samples < 1:
             raise ValueError("individuals_nodes must have at least one sample")
@@ -107,7 +101,7 @@ def iter_field(self, field_name, shape, start, stop):
 
     def iter_alleles_and_genotypes(self, start, stop, shape, num_alleles):
         # All genotypes in tskit are considered phased
-        phased = np.full(shape[:-1], self.is_phased, dtype=bool)
+        phased = np.ones(shape[:-1], dtype=bool)
 
         for variant in self.ts.variants(
             isolated_as_missing=self.isolated_as_missing,

tests/test_simulated_data.py

@@ -1,9 +1,9 @@
 import msprime
+import numpy as np
 import numpy.testing as nt
 import pysam
 import pytest
 import sgkit as sg
-import numpy as np
 
 from bio2zarr import vcf as vcf_mod
 
@@ -35,7 +35,8 @@ def assert_ts_ds_equal(ts, ds, ploidy=1):
     )
     nt.assert_array_equal(
         ds.call_genotype_phased.values,
-        np.ones((ts.num_sites, ts.num_individuals), dtype=bool))
+        np.ones((ts.num_sites, ts.num_individuals), dtype=bool),
+    )
     nt.assert_equal(ds.variant_allele[:, 0].values, "A")
     nt.assert_equal(ds.variant_allele[:, 1].values, "T")
     nt.assert_equal(ds.variant_position, ts.sites_position)

tests/test_tskit.py

@@ -143,7 +143,7 @@ def test_phased(self, conversion):
         phased = zroot["call_genotype_phased"][:]
         assert phased.shape == (3, 4)
         assert phased.dtype == "bool"
-        assert np.all(~phased)
+        assert np.all(phased)
 
     def test_contig_id(self, conversion):
         ts, zroot = conversion
@@ -470,7 +470,17 @@ def test_genotype_dtype_i4(self, tmp_path):
 
 @pytest.mark.parametrize(
     "ts",
-    [add_mutations(msprime.sim_ancestry(2, sequence_length=10, random_seed=42))],
+    [
+        add_mutations(
+            msprime.sim_ancestry(4, ploidy=1, sequence_length=10, random_seed=42)
+        ),
+        add_mutations(
+            msprime.sim_ancestry(2, ploidy=2, sequence_length=10, random_seed=42)
+        ),
+        add_mutations(
+            msprime.sim_ancestry(3, ploidy=12, sequence_length=10, random_seed=142)
+        ),
+    ],
 )
 def test_against_tskit_vcf_output(ts, tmp_path):
     vcf_path = tmp_path / "ts.vcf"