Bayesian models with dominance effects for genomic evaluation of quantitative traits

Area of Science:

• Quantitative genetics
• Animal and plant breeding
• Statistical genomics

Background:

• Genomic selection (GS) uses genome-wide markers to predict breeding values (BV), accelerating genetic gain in breeding programs.
• Current GS primarily focuses on additive genetic effects, potentially overlooking the impact of dominance.
• Dominance effects can be significant even with small dominance variance, influencing quantitative trait loci (QTLs).

Purpose of the Study:

• To introduce a hierarchical Bayesian model for genomic selection that incorporates dominance effects.
• To compare submodels for predicting breeding values, dominance deviations, and genotypic values.
• To evaluate the impact of dominance on prediction accuracy and its utility in mate selection.

Main Methods:

• Developed a general hierarchical Bayesian model for genomic selection accounting for dominance.
• Proposed and compared several submodels differing in additive-dominance effect dependency.
• Utilized stochastic simulation to assess prediction abilities for genomic BV, dominance deviations, and genotypic values (GV).

Main Results:

• Inclusion of dominance effects improved genotypic value (GV) accuracy by ~17% and genomic breeding value (BV) accuracy by ~2% in offspring.
• Dominance effects slowed the decline in prediction accuracy across generations.
• Accurate estimation of genotypic values (GV) was achieved, facilitating mate selection.

Conclusions:

• A hierarchical Bayesian model effectively incorporates dominance into genomic selection.
• Accounting for dominance enhances prediction accuracy for genotypic values and breeding values.
• This model provides a valuable tool for optimizing mate selection strategies in breeding programs.