PREDICTION OF OFFSPRING PROBABILITIES FOR OBSERVED HAPLOTYPE VARIANTS USING PARSIMONY- AND PROBABILITY-BASED METHODS

Abstract

Offspring probabilities for a cattle breeding population were calculated from historical genotypic data on closely-linked single nucleotide polymorphisms (SNPs) using a “haplotype-centric” approach. The number of haplotypes and their corresponding frequencies were estimated using manual parsimony methods and the probability-based methods of HAPLOVIEW (ver. 3.32) and PHASE (ver. 2.1). All methods identified the same set of haplotypes in the population. The Bayes theorem was applied on calculated haplotype frequencies to determine probable haplotypes and their corresponding frequencies for cases of incomplete genotype information (i.e. two out of six loci genotyped), with the assumption of Hardy-Weinberg equilibrium and the absence of recombination. The most probable haplotype frequencies for each incomplete genotype allowed the prediction of offspring probabilities for all possible crosses between individuals. Results show that the minimal set of haplotypes in a population can be determined by different methods. Moreover, the true haplotype of an individual can be predicted even when only a fraction of the SNPs was genotyped by applying Bayesian statistics on the known haplotype frequencies in the population.