Title: "Bayesian Shrinkage Mapping of Quantitative Trait Loci"
Speaker: Professor Shizhong Xu, Department of Botany and Plant Sciences, University of California, Riverside, CA
Place: Mechanical Engineering (ME) 161; November 18, 2008, Tuesday, 4:30pm

Abstract

The purpose of QTL mapping is to use a marker map to infer a finite number of QTL. The number of markers is much more than the number of QTL. Therefore, QTL mapping has long been considered as a model (variable) selection problem. The modern molecular technology has allowed us to generate the amount of marker data that can easily saturate the entire genome of interested species. The saturated marker map makes the model selection problem more difficult than ever. Therefore, new statistical methods and computation algorithms become extremely important in extracting the maximum amount of information from data and separating signals from noise. Bayesian shrinkage analysis is a method alternative to model selection. Instead of selecting a finite number of markers that are strongly associated with the phenotype of a quantitative trait, we actually use all markers without selection to evaluate every segment of the genome simultaneously. The number of QTL is no longer relevant; instead, genome locations that are heavily hit by significant QTL are more important. The Bayesian shrinkage method is combined with the recently popularized LASSO (least absolute shrinkage and selection operator) method to generate a new method called Bayesian LASSO. This hybrid method is more efficient than either method. Applications of the new methods are demonstrated using both simulated and real data.

Associated Reading:

Xu, S. 2003. Estimating polygenic effects using markers of the entire genome. Genetics 163:789-801.

Xu, S. 2007. An empirical Bayes method for estimating epistatic effects of quantitative trait loci. Biometrics 63:513-521.

Yi, N. J. and S. H. Xu. 2008. Bayesian LASSO for quantitative trait loci mapping. Genetics 179:1045-1055.





Click here for a full schedule of BIOINFORMATICS SEMINARS, past and present.