WIP

src/compare_vcfs.py

@@ -69,14 +69,19 @@ def make_compatible_genotypes(ds1, ds2):
     """
     Make `ds2` compatible with `ds1` by remapping genotypes.
 
+    Definition of compatibility:
+    1. `ds1` and `ds2` have the same number of samples.
+    2. `ds1` and `ds2` have the same ploidy.
+    3. `ds1` and `ds2` have the same number of variable sites.
+    4. `ds1` and `ds2` have the same allele list at each site.
+
     :param xarray.Dataset ds1: sgkit-style dataset
     :param xarray.Dataset ds2: sgkit-style dataset
-    :return: Remapped `ds2`
-    :rtype: xarray.Dataset
+    :return: Compatible `ds1` and `ds2`
+    :rtype: tuple(xarray.Dataset, xarray.Dataset)
     """
     remapped_ds2_genotype = remap_genotypes(ds1, ds2)
-    remapped_ds2_genotype = xr.DataArray(remapped_ds2_genotype, dims=('variants', 'samples'))
-    ds1_reduced = None
-    ds2_reduced = None
-    remapped_ds2 = ds2.assign(call_genotype=remapped_ds2_genotype)
-    return remapped_ds2
+    common_site_pos = remapped_ds2_genotype.variant_position
+    ds1_subset = None
+    ds2_subset = None
+    return (ds1_subset, ds2_subset)