Generating consistent genotypic configurations for multi-allelic loci and large complex pedigrees

Markov chain Monte Carlo techniques have been applied to many different genetic analysis problems. Gibbs sampling in particular has been widely used because of its simplicity and because it can be extended to arbitrarily complex pedigrees and genetic models (albeit with modifications for multi-allelic loci). Gibbs sampling requires an initial genotypic configuration, consistent with observed data, the generation of which is not trivial with large complex pedigrees and multi-allelic loci. A method to generate feasible genotype configurations in these circumstances using a combination of peeling and genotype elimination is described. The method is illustrated using two complex multi-generation pedigrees, one real and one simulated, each partially typed for one highly polymorphic marker locus.