Statistical Methods for Relatedness Inference in Admixed Populations
April 10, 2012 @ 3:30 p.m. - 4:30 pm
Timothy Thornton, Ph.D., Department of Biostatistics, University of Washington
Location: 701 Blockley Hall
Abstract: Genome-wide association studies (GWAS) are commonly used for the mapping of genetic loci that influence complex traits. A problem that is often encountered in GWAS is that of identifying cryptic relatedness and population structure (ancestry differences among sample individuals), as it is well known that failure to appropriately account for both pedigree and population structure can lead to spurious association. Genetic models used to identify related individuals in large-scale genetic data, however, often make simplifying assumptions – either random mating or simple structures. In reality, human populations do not mate at random nor are there simple endogamous subgroups. For example, in the U.S., the amount of intercontinental admixture and inter-mating between racial/ethnic groups is increasing, but at the same time there is evidence of ancestry-related assortative mating within ethnic groups. In these circumstances, it is necessary to devise statistical methods that account for the diverse genomes of the sample individuals and are robust in the presence of a variety of complex, ancestry-related mating patterns. We propose a new method for relatedness inference in the presence of admixture and ancestry-related assortative mating. We apply the method to African American and Hispanic samples from the Women's Health Initiative study where hundreds of pairs of cryptically related individuals are identified.