Merge pull request #557 from hyanwong/new-map-mutations

Switch to the (more correct) new map_mutations code

requirements/CI-tests-complete/requirements.txt

@@ -16,5 +16,5 @@ pytest-xdist==2.2.1
 seaborn==0.11.1
 sortedcontainers==2.3.0
 tqdm==4.60.0
-tskit==0.3.5
+tskit==0.3.7
 zarr==2.8.1

requirements/CI-tests-pip/requirements.txt

@@ -10,4 +10,4 @@ pytest-xdist==2.2.1
 seaborn==0.11.1
 sortedcontainers==2.3.0
 tqdm==4.61.1
-tskit==0.3.6
+tskit==0.3.7

setup.py

@@ -81,7 +81,7 @@ def finalize_options(self):
         "tqdm",
         "humanize",
         "daiquiri",
-        "tskit>=0.3.3",
+        "tskit>=0.3.7",
         "numcodecs>=0.6",
         "zarr>=2.2",
         "lmdb",

tsinfer/inference.py

@@ -706,7 +706,7 @@ def insert_missing_sites(
     for variant in sample_data.variants(sites=new_sd_sites):
         site = variant.site
         pos = site.position
-        anc_state = site.ancestral_state
+        anc_state = site.ancestral_state  # NB: None means an unknown ancestral state
         anc_value = 0  # sample_data files always have 0 as the ancestral allele idx
         G = variant.genotypes[sample_id_map]
         # We can't perform parsimony inference if all sites are missing, and there's no
@@ -725,11 +725,9 @@ def insert_missing_sites(
         else:
             while tree.interval[1] <= pos:
                 tree = next(trees)
-            inferred_anc_state, mapped_mutations = tree.map_mutations(
-                G, variant.alleles
+            anc_state, mapped_mutations = tree.map_mutations(
+                G, variant.alleles, ancestral_state=anc_state
             )
-            if anc_state is None:
-                anc_state = inferred_anc_state
             metadata = _update_site_metadata(
                 site.metadata, inference_type=constants.INFERENCE_PARSIMONY
             )
@@ -741,15 +739,6 @@ def insert_missing_sites(
                 metadata=metadata,
             )
             mut_map = {tskit.NULL: tskit.NULL}
-            if anc_state != inferred_anc_state:
-                # Need to add an extra mutation above the root to switch ancestral state
-                for root in tree.roots:
-                    mut_map[tskit.NULL] = tables.mutations.add_row(
-                        site=new_site_id,
-                        node=root,
-                        derived_state=inferred_anc_state,
-                        parent=tskit.NULL,
-                    )
             for i, mutation in enumerate(mapped_mutations):
                 mut_map[i] = tables.mutations.add_row(
                     site=new_site_id,