Factors affecting GEBV accuracy with single-step Bayesian models

Area of Science:

• Animal Breeding and Genetics
• Quantitative Genetics
• Bioinformatics

Background:

• Genomic selection (GS) utilizes genomic information for predicting breeding values.
• Single-step genomic best linear unbiased prediction (SSGBLUP) integrates all related individuals in genomic evaluation.
• Understanding factors affecting genomic estimated breeding value (GEBV) accuracy in single-step analysis is crucial for genetic improvement.

Purpose of the Study:

• To explore the components of GEBV accuracy in single-step Bayesian analysis.
• To investigate the influence of population structure and relationships between training and validation populations on GEBV accuracy.
• To compare the performance of single-step genomic best linear unbiased prediction (GBLUP; SSGBLUP), single-step BayesA (SS-BayesA), and single-step BayesB (SS-BayesB) models.

Main Methods:

• A simulation study was conducted with three scenarios of quantitative trait loci (QTL) numbers (5, 50, and 500).
• Three models were implemented: SSGBLUP, SS-BayesA, and SS-BayesB.
• GEBV accuracy was assessed based on relationships between training and validation populations and linkage disequilibrium (LD) between markers and QTL.

Main Results:

• GEBV accuracy was more sensitive to population relationships for ungenotyped than genotyped animals.
• SS-BayesA and SS-BayesB outperformed SSGBLUP when traits were controlled by 5 or 50 QTL.
• SS-BayesA demonstrated the most robust and efficient performance across all QTL scenarios, while SS-BayesB showed lower accuracy with 500 QTL.

Conclusions:

• Population structure and marker-QTL relationships significantly contribute to GEBV accuracy in single-step analyses.
• Single-step Bayesian models, particularly SS-BayesA, offer advantages over SSGBLUP, especially for traits influenced by a limited number of QTL.
• The choice of model impacts GEBV accuracy, highlighting the importance of considering the genetic architecture of traits in genomic selection strategies.