Efficient single-step genomic evaluation for a multibreed beef cattle population having many genotyped animals

Area of Science:

• Animal Breeding and Genetics
• Quantitative Genetics
• Bioinformatics

Background:

• Single-step genomic best linear unbiased prediction (ssGBLUP) is a standard method for genetic evaluations.
• Computational demands of ssGBLUP can be high, especially for large datasets.
• Efficient computational methods are crucial for accurate and timely genetic predictions.

Purpose of the Study:

• To introduce and evaluate ssGTBLUP, an equivalent computational approach to ssGBLUP.
• To compare the performance of ssGTBLUP and its approximation ssGTBLUP(p) against ssGBLUP and the APY method.
• To assess the accuracy and computational efficiency of these methods in a large-scale beef cattle genetic evaluation.

Main Methods:

• Formulated ssGTBLUP by modifying the genomic relationship matrix for efficient inversion.
• Developed ssGTBLUP(p) using eigendecomposition to reduce computational complexity.
• Compared ssGTBLUP, ssGTBLUP(p), ssGBLUP, and APY methods on Irish beef carcass conformation data.
• Utilized a large pedigree (13.3 million animals) and marker data (54,620 markers from 163,277 animals).

Main Results:

• ssGTBLUP achieved significant reductions in computation time per iteration compared to ssGBLUP.
• ssGTBLUP(p) showed high correlations (0.992-1.000) with ssGBLUP breeding values for various approximation levels (p=95-99).
• APY methods showed lower correlations with ssGBLUP, especially with smaller core populations (APY10K: 0.899-0.967).
• Computing times per iteration were substantially reduced with ssGTBLUP(p) and APY methods.

Conclusions:

• ssGTBLUP provides a computationally efficient alternative to ssGBLUP.
• The ssGTBLUP(p) approach offers a practical and accurate approximation for large-scale genetic evaluations.
• ssGTBLUP(p) significantly reduces computational burden while maintaining high accuracy of breeding values.