BayesR3 enables fast MCMC blocked processing for largescale multi-trait genomic prediction and QTN mapping analysis.

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genomics

Background:

• Bayesian methods, including BayesR, are used for genetic prediction from genotypes like Single Nucleotide Polymorphisms (SNPs).
• Markov Chain Monte Carlo (MCMC) processes, commonly used in these methods, are computationally intensive and slow.
• Efficient computation is crucial for large-scale genomic prediction and fine-mapping.

Purpose of the Study:

• To introduce a computationally efficient blocked Gibbs sampling strategy for Bayesian SNP effect estimation.
• To reduce the computational time associated with MCMC in genetic prediction models.
• To enhance the precision of mapping genetic variants for complex traits using large-scale genomic data.

Main Methods:

• Implemented a blocked Gibbs sampling approach for estimating SNP effects within a Markov Chain Monte Carlo (MCMC) framework.
• Developed the BayesR3 model, a Bayesian MCMC mixed-effects genetic model, incorporating the blocked Gibbs sampling strategy.
• Validated the method with simulated data and applied it to empirical dairy cattle data using high-dimensional omics data (milk mid-infrared spectra).

Main Results:

• The blocked Gibbs sampling method significantly reduced computational time compared to standard MCMC.
• The BayesR3 model demonstrated increased precision in mapping variants associated with milk, fat, and protein yields.
• The approach proved effective for large-scale genomic prediction and fine-mapping, outperforming univariate analysis.

Conclusions:

• Blocked Gibbs sampling offers a computationally efficient alternative for Bayesian genomic prediction and fine-mapping.
• The BayesR3 model enhances the accuracy and precision of identifying genetic variants influencing complex traits.
• This method is valuable for analyzing large-scale 'omics' data in livestock for genetic improvement.