Bayesian methods for jointly estimating genomic breeding values of one continuous and one threshold trait

Area of Science:

• Quantitative genetics
• Animal and plant breeding

Background:

• Genomic selection is vital for breeding, typically using single-trait models.
• Multi-trait models offer improved genomic prediction accuracy by leveraging correlated trait information.
• Joint genomic prediction for continuous and threshold traits is underexplored.

Purpose of the Study:

• To develop a novel multi-trait model for joint genomic prediction of continuous and threshold traits.
• To evaluate the proposed method's performance against single-trait models.
• To investigate factors influencing the joint prediction accuracy.

Main Methods:

• Development of the LT-BayesCπ method based on a linear-threshold model.
• Derivation of computational procedures using the Markov Chain Monte Carlo algorithm.
• Simulation studies to compare LT-BayesCπ with BayesCπ (continuous) and BayesTCπ (threshold).

Main Results:

• LT-BayesCπ significantly increased genomic prediction accuracy for threshold traits compared to single-trait prediction.
• Accuracy for continuous traits remained comparable to single-trait prediction.
• The method's performance was evaluated across various scenarios.

Conclusions:

• The LT-BayesCπ method provides a robust approach for joint genomic prediction of one continuous and one threshold trait.
• This method enhances accuracy for threshold traits, offering a valuable tool for breeding programs.
• LT-BayesCπ represents a significant advancement in multi-trait genomic evaluation.