Using Markov chain Monte Carlo for multipoint linkage analysis: Improved estimates of lod scores

The calculation of exact likelihoods from pedigree data is limited to datasets containing either a small number of meioses, or a small number of linked genetic loci. In particular, the computation of likelihoods from data collected at multiple loci on large, extended pedigrees is infeasable. We perform multipoint linkage analysis on such datasets by estimating ratios of these otherwise intractable likelihoods. Using Markov chain Monte Carlo (MCMC) and a robust system of weights, we transform the set of ratio estimates into an estimate of the lod score curve. We demonstrate our ability to estimate accurately multipoint lod scores in cases where exact computation is fe<;tsible. In a situation well beyond the scope of exact computation, we show the agreement between our estimates and those obtained from another MCMC method. Our method provides a means for achieving reliable, accurate analysis of data collected at multiple linked loci on large, extended pedigrees.