Gene identity by descent (ibd) underlies all genetically mediated similarities among relatives. The ibd graph, defined among observed individuals and across the genome, specifies the segments of genome shared ibd among these individuals. If pedigree relationships among individuals are known, relatively sparse genetic marker data serve to estimate the latent ibd graph. Once the ibd graph is known, analyses of quantitative trait data may be carried out conditionally on the ibd graph, and the pedigree relationships and genetic marker data are no longer relevant. More remote relationships among members of different pedigrees are usually not known. The ibd resulting from these remote relationships can be estimated using denser genetic marker data and a population-genetic based ibd model. Specifically, we estimate the ibd graphs among the four genomes of pairs of individuals. Algorithms to merge the ibd graphs inferred within and among pedigrees provide a combined ibd graph. Using this combined graph for subsequent trait-data analysis increases both the power and the resolution of mapping of genes contributing to complex quantitative traits.
Keywords: Gene coancestry; latent variables; graphical models; genetic linkage
Biography: Elizabeth Thompson has been a Professor of Statistics at the University of Washington, Seattle, since 1985. She is also an adjunct professor in the departments of Biostatistics and of Genome Sciences at the University of Washington. Dr. Thompson's research is in the development of methods for model-based likelihood inference from genetic data, particularly from data observed on large and complex pedigree structures both of humans and of other species, and including inference of relationships and genome coancestry among individuals and among populations.