Statistical power of an exact test of Hardy-Weinberg proportions of genotypic data at a multiallelic locus.

A computer algorithm for numerical evaluation of the statistical power of an exact test of Hardy-Weinberg genotypic proportions (HWP), developed here, indicates that the power is dependent on the number of segregating alleles as well as allele frequencies. While low levels of departure from the null hypothesis are difficult to detect from single-locus data, should such deviation be due to population substructuring, multiple loci, at each of which the number of segregating alleles is large (as seen with hypervariable loci), may easily detect even low levels of departure from HWP. Undetected small levels of departure may still provide conservative estimates of genotype frequencies from allele frequency data, following the current practice in forensic genetics.