Genomic prediction for meat and carcass traits in Nellore cattle using a Markov blanket algorithm

Area of Science:

• Animal Genetics and Breeding
• Genomic Prediction
• Quantitative Genetics

Background:

• Genomic selection (GS) is crucial for improving livestock traits.
• Accurate genomic prediction requires selecting relevant genetic markers.
• The Markov blanket algorithm offers a method for identifying informative single nucleotide polymorphism (SNP) markers.

Purpose of the Study:

• To evaluate the advantage of preselecting SNP markers using the Markov blanket algorithm.
• To assess the impact on genomic prediction accuracy for carcass and meat quality traits in Nellore cattle.
• To compare prediction accuracies of different Bayesian genomic regression models.

Main Methods:

• Utilized data from Nellore cattle for rib eye area (REA), back fat thickness (BF), rump fat (RF), and Warner-Bratzler shear force (WBSF).
• Genotyped animals with low-density SNP panels (30k) and imputed to 777k SNPs.
• Compared four Bayesian genomic regression models (Bayes A, B, Cπ, BRR) using five-fold cross-validation.
• Applied the Markov blanket algorithm to preselect informative SNP markers.

Main Results:

• Prediction accuracies for REA, BF, and RF were similar across Bayesian Alphabet models (0.75-0.95).
• Bayes B showed higher predictive ability for WBSF (0.47) compared to other methods (0.39-0.42).
• While accuracies with Markov blanket subsets were lower than using all SNPs, the relative gain for WBSF was less than 13%.

Conclusions:

• Preselecting informative SNP markers via Markov blanket may capture significant genetic variance for WBSF.
• Developing customized, low-density SNP arrays using Markov blanket could be cost-effective for genomic selection.
• This approach can enhance the scale and efficiency of genomic selection in cattle breeding programs.