Genomic prediction with whole-genome sequence data in intensely selected pig lines

Area of Science:

• Animal Genetics
• Genomic Prediction
• Quantitative Genetics

Background:

• Simulations suggested whole-genome sequence data (WGS) could enhance genomic prediction accuracy within and across breeds.
• Empirical evidence has been inconsistent, necessitating large, diverse datasets for accurate allele substitution effect estimation.

Purpose of the Study:

• To assess the benefits of using WGS versus commercial marker arrays for genomic prediction in pigs.
• To identify specific scenarios where WGS provides the greatest advantage in intensely selected lines.

Main Methods:

• Sequenced 6931 pigs across seven commercial lines; imputed genotypes for 396,100 individuals.
• Employed BayesR for genomic prediction of eight complex traits using marker array data or WGS variants selected via association tests.

Main Results:

• Preselected WGS variants showed inconsistent prediction accuracy improvements across traits and lines.
• WGS benefits were most pronounced with large training sets (approx. 80k individuals), augmenting marker arrays with significant WGS variants, yielding average within-line prediction accuracy improvements of 0.025.
• Multi-line training sets showed potential for 0.04 accuracy improvements.

Conclusions:

• WGS currently offers limited accuracy gains over marker arrays for genomic prediction in intensely selected pig lines.
• Larger training datasets and optimized WGS data analysis pipelines could increase benefits.
• The cost-benefit of WGS for genomic prediction requires case-by-case evaluation.