The method begins by choosing initial guesses for the haplotypes in – by default, we choose haplotypes that are consistent with the observed genotype data but phased at random. We then perform a number of MCMC iterations. Each iteration updates every study individual i (in some arbitrary order) in two steps:
