Software Development for Deterministic Prediction of Selection Response in Livestock Breeding Programs Using Genomic Information

Theory to predict selection response in traditional livestock breeding programs has been well developed, validated and implemented in software in the past decades, for example in SelAction (Rutten et al. 2002), which has been successful as a tool to predict selection response in traditional livestock breeding programs for a wide range of population structures and selection strategies. This software used standard quantitative genetics theory and selection index theory to develop deterministic recursive equations, which model changes of trait means and variance-covariance structures to predict asymptotic response to multiple trait selection using best linear unbiased prediction (BLUP) estimated breeding values (EBV). Nowadays genetic improvement can further be enhanced by genomic predictions, which provide more accurate estimates of breeding values of animals in their earlier life and can improve the efficiency of breeding programs. While statistical methods to estimate genomic breeding values are now widely available, optimizing the use of genomics in practical livestock breeding programs is limited due to the lack of computer software that implements available theories. We're hereby to present a computer program that extends SelAction. Genomic information is included as the average phenotype of groups of individuals with both genotypic and phenotypic information following Wientjes et al. (2016). The heterogeneity of genomic information is considered in terms of the degree of relationship between selection candidates and the individuals that are both genotyped and phenotyped (van der Werf et al., 2015). This software can be used by breeders to reliably compare alternative breeding programs and for investment decisions for breeding programs that include genomic information. Funded by USDA-NIFA grant #2017-67015-26299.