For each individual, we simulate genotype data at 10,000 uncorrelated SNPs not associated with disease, but used to calculate IBS relationships. For each SNP, we simulate a minor allele frequency, q, in the interval [0.05, 0.5] in the case-control source population, and generate genotypes assuming Hardy-Weinberg equilibrium, irrespective of disease status. We then generate the allele frequency for the same SNP, in each external cohort in turn, using the Balding-Nichols model [Balding and Nichols, 1995]. Specifically, the allele frequency is simulated from a Beta(a,b) distribution, where a = q(1 − FST)/FST and b = (1−q)(1-FST)/FST [Devlin et al., 2001]. Again, genotypes are simulated with this allele frequency under the assumption of Hardy-Weinberg equilibrium.
