Comparative Study of Single-Trait and Multi-Trait Genomic Prediction Models

Area of Science:

• Quantitative genetics
• Animal breeding
• Genomic selection

Background:

• Conventional genomic selection models analyze traits individually, overlooking complex genetic interactions.
• Multi-trait models incorporate genetic correlations to improve the accuracy of breeding value estimation.

Purpose of the Study:

• To evaluate the breeding advantages of multi-trait genomic best linear unbiased prediction (GBLUP) models.
• To assess model performance across varying population sizes and genetic correlation levels.
• To investigate the impact of heritability on multi-trait model benefits.

Main Methods:

• Simulations using 50K chip data from 5000 individuals.
• Evaluation of multi-trait GBLUP against single-trait models.
• Analysis under different heritability scenarios (equal and differing) and genetic correlation levels (low, medium, high).

Main Results:

• Multi-trait GBLUP consistently outperformed single-trait models in equal heritability scenarios, with gains increasing alongside heritability.
• Improvements ranged from 0.3% to 4.1% with a reference population of 4500.
• Low heritability traits showed minimal gains (≤ 0.1%) regardless of genetic correlation.
• Benefits varied in differing heritability scenarios, notably enhancing low-heritability traits when paired with high-heritability ones.
• Modeling time increased as genetic correlation decreased.

Conclusions:

• Multi-trait models enhance breeding accuracy but demand increased computational resources and modeling time.
• Tailored breeding strategies are recommended to balance efficiency and accuracy based on phenotypes and genetic backgrounds.