check allele codes before trying to write pgen

haptools/data/genotypes.py

@@ -1590,6 +1590,7 @@ def write(self):
                 end = start + chunks
                 if end > len(self.variants):
                     end = len(self.variants)
+                self.log.debug(f"Writing variant #{start} through variant #{end}")
                 size = end - start
                 try:
                     missing = np.ascontiguousarray(
@@ -1606,6 +1607,8 @@ def write(self):
                         "You don't have enough memory to write these genotypes! Try"
                         " specifying a value to the chunk_size parameter, instead"
                     ) from e
+                if not np.all(allele_cts <= max_allele_ct):
+                    raise ValueError("Some variants have more alleles than expected")
                 # convert any missing genotypes to -9
                 subset_data[missing] = -9
                 # finally, append the genotypes to the PGEN file

haptools/sim_genotype.py

@@ -111,6 +111,7 @@ def output_vcf(
         vcf.read()
     else:
         vcf.read(region=f"{region['chr']}:{region['start']}-{region['end']}")
+    vcf.check_missing()
 
     log.debug(f"Read in variants from {variant_file}")