The overall objective is ttie development of methods for the enhanced detection and resolution of genes contributing to complex quantitative genetic traits observed in individuals not known to be related. The approach will be through using dense SNP marker data for the detection and estimation of segments of gene identity by descent (ibd) shared among sets of individuals. Locus-specific inferred ibd among individuals will be analyzed in conjunction with their phenotypic similarities and differences, in order to detect and resolve causal loci. We will develop and assess hidden Markov models (HMM) and methods for detection of ibd genome segments between pairs of members of populations from dense SNP data or sequence variants. We will assess the effects on performance of our methods of linkage disequilibrium, data error and copynumber variants, and the efficacy of prior haplotype imputation, data cleaning, and screening for regions of allelic similarity. We will extend our models and methods to the inference of ibd among larger sets of chromosomes using both HMM and coalescent models, and develop Markov chain Monte Carlo methods for sampling of ibd genome segments, conditional on dense SNP marker or sequence variant data in candidate gene regions. We will develop and assess methods for analyzing trait data on individuals conditional on the patterns of ibd genome segments inferred among them, by assessing location-specific levels and regional chromosomal extent of ibd segments among sampled chromosomes in relation to quantitative trait values. We will assess our methods and compare with alternative approaches, by first testing methods in simulated population structures, where latent ibd is known, but in which founder haplotypes are provided by real-data population samples. Then, in real data sets available to us, where latent ibd is unknown, we will compare results of our methods with those of other approaches developed both within the P01 group and elsewhere. We will develop software implementing our methods, and document, distribute and support this software.
Showing the most recent 10 out of 152 publications
