Title:Locally epistatic genomic relationship matrices for genomic association, prediction and selection

Abstract:As the amount and complexity of genetic information increases it is necessary that we explore some efficient ways of handling these data. This study takes the "divide and conquer" approach for analyzing high dimensional genomic data. Our aims include reducing the dimensionality of the problem that has to be dealt one at a time, improving the performance and interpretability of the models. We propose using the inherent structures in the genome; to divide the bigger problem into manageable parts. In plant and animal breeding studies a distinction is made between the commercial value (additive + epistatic genetic effects) and the breeding value (additive genetic effects) of an individual since it is expected that some of the epistatic genetic effects will be lost due to recombination. In this paper, we argue that the breeder can take advantage of some of the epistatic marker effects in regions of low recombination. The models introduced here aim to estimate local epistatic line heritability by using the genetic map information and combine the local additive and epistatic effects. To this end, we have used semi-parametric mixed models with multiple local genomic relationship matrices with hierarchical testing designs and lasso post-processing for sparsity in the final model and speed. Our models produce good predictive performance along with genetic association information.