Accuracy of genomic prediction of maternal traits in pigs using Bayesian variable selection methods

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Reproductive Biology

Background:

• Genomic prediction aims to improve selection accuracy for complex traits.
• Maternal traits in sows are crucial for production but often have low heritability and are difficult to record.
• Genomic best linear unbiased prediction (GBLUP) assumes equal SNP effects, while Bayesian methods can account for varying SNP effects and animal relationships.

Purpose of the Study:

• To compare the accuracy of three genomic prediction methods: GBLUP, BayesC, and BayesGC.
• To evaluate the impact of different priors on Bayesian prediction methods.
• To assess prediction accuracy for six maternal traits in Landrace sows using a 660K SNP panel.

Main Methods:

• Genomic prediction using GBLUP, BayesC, and BayesGC methods.
• Investigation of different prior distributions for Bayesian genomic prediction.
• Application to six maternal traits in Landrace sows with a 660K SNP panel.

Main Results:

• BayesGC generally exhibited higher prediction accuracy across several maternal traits.
• For piglet mortality within 3 weeks, BayesGC achieved up to 9.2% higher accuracy compared to other methods.
• In some instances, GBLUP or BayesC matched the accuracy of BayesGC for specific traits.
• Differences in accuracy between methods were not significant for many of the evaluated traits.

Conclusions:

• BayesGC is a promising method for genomic prediction of maternal traits in sows, offering improved accuracy for certain traits like piglet mortality.
• The choice of genomic prediction method can influence accuracy, highlighting the importance of considering genetic architecture.
• Further research into Bayesian methods and prior selection may enhance genomic selection efficiency in pig breeding programs.