Comparison between genomic predictions using daughter yield deviation and conventional estimated breeding value as...

Area of Science:

• Animal Breeding and Genetics
• Quantitative Genetics
• Genomic Selection

Background:

• Genomic selection (GS) is crucial for improving livestock breeding.
• Accurate prediction of breeding values relies on appropriate response variables.
• Estimated Breeding Values (EBV) and Daughter Yield Deviations (DYD) are common response variables for GS.

Purpose of the Study:

• To compare the accuracy of genomic predictions using EBV versus DYD as response variables.
• To evaluate the impact of different simulation scenarios (heritability, daughter numbers, genotyped sire proportion) on prediction accuracy.
• To assess the performance of EBV and DYD under various statistical models (Bayesian common prior, Bayesian mixture prior, GBLUP).

Main Methods:

• Simulated genetic data with varying heritability, daughter numbers per sire, and proportion of genotyped sires.
• Employed Bayesian common prior and Bayesian mixture prior models for marker effect variance.
• Utilized a linear mixed model (GBLUP) for genomic prediction.
• Calculated genomic estimated breeding value (GEBV) reliabilities and correlation with parent average (PA).

Main Results:

• EBV approach showed similar or slightly superior GEBV reliabilities compared to DYD across most simulated scenarios.
• EBV was notably superior in scenarios with unequal daughter numbers and fewer genotyped sires.
• The EBV approach resulted in a lower correlation between GEBV and PA, indicating a greater potential gain from including PA in selection indices.

Conclusions:

• Conventional EBV is a reliable and effective response variable for genomic prediction, comparable or superior to DYD.
• EBV offers advantages in specific scenarios, particularly when data is unbalanced or genotyped sire numbers are limited.
• The choice of response variable can influence the integration of genomic and pedigree information in selection indices.