Approximate Computation of Probability Functions for Pedigree Analysis

Abstract

Approximation by factorization of R functions is considered as a way of making the peeling method for calculation of probability functions more efficient and applicable to larger and more complex pedigrees. We show that, in some situations, the method is not approximate but a simplified way of calculating the true probability functions. The entropy function is used for a measure of the quality of factorization. The method is illustrated by an analysis of the origin of the B allele in the ABO bloodtype system on Tristan da Cunha.