Abstract

Multiparent populations (MPP) have become popular resources for complex trait mapping because of their wider allelic diversity, and larger population size compared to traditional two-way recombinant inbred (RI) strains. In Mice, the Collaborative Cross (CC) is one of the most popular MPP and is derived from eight genetically diverse inbred founder strains. The strategy of generating recombinant inbred intercrosses (RIX) from MPP in general and the CC in particular can produce a large number of completely reproducible heterozygote genomes that better represent the (outbred) human population. Since both maternal and paternal haplotypes of each RIX are readily available, RIX is a powerful resource for studying both standing genetic and epigenetic variations of complex traits, in particular, the Parent-of-Origin (PoO) effects, an important contributor to many complex traits. Furthermore, most complex traits are affected by more than one genes where multiple QTL mapping could be more advantageous. In this paper, for MPP-RIX data but taking CC-RIX as a working example, we propose a general Bayesian variable selection procedure to simultaneously search for multiple genes with founder allelic effects and PoO effects. The proposed model respects the complex relationship among RIX samples and the performance of the proposed method is examined by extensive simulations.