Genomic prediction from observed and imputed high-density ovine genotypes

Area of Science:

• Animal genetics
• Quantitative genetics
• Genomic selection

Background:

• High-density (HD) marker genotypes are anticipated to enhance genomic prediction accuracy, especially in heterogeneous multi-breed and crossbred populations like sheep and beef cattle.
• HD genotypes strengthen linkage disequilibrium between single nucleotide polymorphisms and quantitative trait loci, crucial for accurate trait prediction.

Purpose of the Study:

• To assess the improvement in genomic prediction accuracy for production traits in Australian sheep breeds using HD genotypes (600k observed and imputed) versus 50k marker genotypes.
• To specifically compare prediction accuracy gains for animals distantly related to the reference population and across different sheep breeds.

Main Methods:

• Genomic best linear unbiased prediction (GBLUP) and a Bayesian approach (BayesR) were employed for prediction.
• A large multi-breed/crossbred sheep reference set was utilized, with prediction accuracy evaluated using the Pearson correlation coefficient.
• Empirical prediction accuracy was assessed for purebred Merino, Border Leicester, Poll Dorset, and White Suffolk sire breeds against progeny test data.

Main Results:

• HD genotypes resulted in a modest average improvement of 2.2% in prediction accuracy for purebred animals across all traits.
• A more substantial increase in prediction accuracy, averaging 5.2%, was observed for animals with low genetic relatedness to the reference set.
• Across-breed prediction showed an improvement ranging from 0.0% to 5.0%, with no significant advantage of BayesR over GBLUP on average.

Conclusions:

• HD genotypes provide a marginal but valuable enhancement to genomic prediction accuracy in sheep, particularly for genetically distant individuals.
• The study highlights the utility of HD markers for improving genomic selection in diverse sheep populations, with implications for breeding programs.
• While BayesR showed no consistent advantage over GBLUP, the use of HD markers generally improved prediction accuracy, especially in challenging scenarios.