The methods also are applicable to data such as microsatellites, where there are more than two alleles at a single site. We use a device of Cavalli-Sforza [2,15], making a “marker” j out of each allele, and then setting C(i,j) to be the number of occurrences of the allele for sample i. We omit the normalization step of Equation 3 for microsatellites, merely subtracting the mean. The normalization has no clear justification for microsatellite data, and results on real data (unpublished) show that it produces worse performance in this case.
