E. A. Thompson - Conditional genome sharing from dense marker maps

With increasing marker data availability and ever-improving genetic maps, localization of the genes contributing to complex traits remains a hard problem. Methods for estimating gene locations are sensitive to trait model assumptions, particularly when multiple markers are analyzed jointly. Robust methods lack power for linkage detection, and localization can be problematic. Data on multiple relatives can provide more power, but valid analysis of data on multiple relatives jointly at multiple marker loci raises severe computational issues.

Markov chain Monte Carlo (MCMC) methods provide realizations of gene identity by descent among pedigree members, conditional on data at multiple marker loci, in situations in which exact computation is infeasible. This estimated gene sharing permits detection and localization of genes contributing to a trait, determination of the pedigrees a trait allele, and inference of gene carriers.