Better formatted error message for ancestral alleles

tsinfer/formats.py

@@ -2405,11 +2405,16 @@ def sites_ancestral_allele(self):
             except IndexError:
                 unknown_alleles[allele] += 1
             ret[i] = allele_index
-        if sum(unknown_alleles.values()) > 0:
+        tot = sum(unknown_alleles.values())
+        if tot > 0:
             warnings.warn(
-                "The following alleles were not found in the variant_allele array "
-                "and will be treated as unknown:\n"
-                f"{unknown_alleles}"
+                "Ancestral alleles were not found in the variant_allele "
+                f"array for {tot} sites ({tot/len(string_allele)*100:.2f}%). "
+                "They are listed below, and will be treated as unknown:\n "
+                + "\n ".join(
+                    f"'{k}': {v} ({v/len(string_allele)*100:.2f}% of sites)"
+                    for k, v in unknown_alleles.items()
+                )
             )
         return ret