Short communication: Genomic prediction using different single-step methods in the Finnish red dairy cattle...

Area of Science:

• Animal Breeding and Genetics
• Quantitative Genetics
• Genomic Selection

Background:

• Single-step genomic prediction models integrate genotyped and non-genotyped animals for advanced genetic evaluations.
• Various single-step models exist, differing in marker effect estimation or direct genomic relationship matrix prediction.

Purpose of the Study:

• Compare classical, single-step genomic BLUP (ssGBLUP), and single-step Bayesian regression (ssBR) models.
• Evaluate model performance for 305-day production traits in Finnish red dairy cattle.
• Assess the impact of a residual polygenic effect on genomic prediction accuracy.

Main Methods:

• Compared a pedigree-based animal model, regular ssGBLUP, ssBR, and Algorithm for Proven and Young (APY) methods.
• Included a residual polygenic effect in single-step models to mitigate prediction inflation.
• Calculated validation reliability using Pearson correlation between GEBV and yield deviation for masked records.

Main Results:

• Single-step methods improved validation reliability by 0.02–0.04 compared to the classical animal model.
• Regular ssGBLUP and ssBR models with a polygenic effect showed comparable results.
• APY methods achieved similar reliabilities to regular ssGBLUP and ssBR.
• ssBR provided exact prediction error variance, unlike ssGBLUP approximations.

Conclusions:

• Single-step genomic prediction models offer enhanced accuracy for livestock genetic evaluations.
• ssBR provides a more precise estimation of prediction error variance, beneficial for large datasets.
• The inclusion of a residual polygenic effect is crucial for reducing inflation in genomic predictions.