Genomic Selection for Gene Introgression

With genomic selection, the loci underlying the trait need not be known a priori. However in many cases, there are robust markers available for desirable genes with significant effects. These markers can be included in models as fixed effects and significantly improve the prediction accuracy over models only using genome-wide markers. We evaluated prediction models incorporating markers linked to important stem rust adult plant resistance (APR) loci as fixed effects along with more commonly used prediction models: Multiple linear regression (MLR), Genomic best linear unbiased prediction (G-BLUP), Bayesian LASSO, and Bayes Cπ. We found the Sr2 region to play an important role in APR in this germplasm. In this study, accuracies from GS models and MLR using only Sr2 linked markers were not significantly different. However, a model using Sr2 linked markers as fixed effects in G-BLUP was the most accurate, and significantly more accurate than MLR using Sr2 linked markers alone. Thus, when large effect loci information is known, targeting these loci with markers and treating these markers as fixed effects is recommended. If introgression of loci from exotic sources is the goal, exotic x elite materials will need to be phenotyped and genotyped for model training in order to be able to accurately predict the performance of exotic x elite materials. Analyses of results from GS experiments for the introgression of exotic germplasm will facilitate a better understanding of the dynamic genome processes that generate and maintain new genetic variation.