Statistical and Computational Issues in Mapping Genes in Animal Populations
Fengxing Du, (Monsanto Company ), email@example.com
Coarse mapping of quantitative trait loci (QTL) in farm animals is commonly performed under simple designs (e.g., line cross and large half-sib families). Tracing inheritance of genes under these designs using linked markers over only a few generations is generally straightforward, but can only achieve coarse mapping. Robust statistical methods have been developed to analyze marker genotype and trait phenotype association under these designs. Complex pedigrees can achieve more experimental power for QTL detection and more accurate QTL parameter estimation, and the framework of analyzing marker data in general pedigrees using mixed models have been developed. However, most animal populations contain a very large number of inbreeding loops, and the estimation of identity by decent probability conditional on linked marker data, that is vital to marker analysis in complex pedigrees, is still a challenge. Currently, approximate methods that only use partial information and Markov chain Monte Carlo based genotype sampling algorithms are being developed to solve this problem. Fine scale mapping of QTL presents additional challenges: it requires a large number of recombinants in small chromosome regions to achieve it. Use of historical recombinants was therefore proposed, by modeling coancestry of a sample conditional on historical assumptions and observed marker data at current generations via coalescence based methods. Other challenges in QTL mapping include model selection (e.g., tightly linked QTL vs. one QTL), interaction between genes, and incorporation of expression profiling data.