Model Selection Techniques for Locating Quantitative Trait Loci: Modifying the Bayesian Information Criterion for Locating Multiple Quantitative Trait Loci in Experimental Crosses

Recent technological advances in genetics pose new challenging problems to statistical methodology. The genetic technique of QTL mapping deals with locating genes that influence a quantitative character of an organism. Within the statistical framework of model selection, existing approaches are adapted and modifed: The search for genetic effects is extended to interactions between genes (i.e. epistatsis), while taking computational performance into account. Robust statistical methods are included in order to cope with quantitative effects that have heavy-tailed distribution or outliers. The performance and applicability of all methods is assessed by extensive simulation studies and the reanalysis of previously published genetic data sets.