Hypothesis testing in genetic linkage analysis via Gibbs sampling

Abstract

Genetic linkage analysis involves estimating parameters in a genetic model in which a genetic trait is regressed on some factors such as polygenic values and environmental effects. Since only phenotypes are observed, hypothesis testing in such cases needs calculation of likelihood function in which one needs to consider all compatible configurations of genotypes. The number of these configurations increases as the size of a pedigree and the number of loci involved increase, Monte Carlo methods play an important role. The existing theory assumes an asymptotic normality for score statistics which is violated on boundary values which is the case in genetic linkage analysis. In this paper, a Markov Chain Monte Carlo approach is proposed to overcome this problem.

Key words: Gibbs sampling, pedigree, linkage analysis, likelihood.