ASSESSMENT OF GENOME-WIDE PREDICTION BY USING BAYESIAN REGULARIZED NEURAL NETWORKS