Genomic Selection for Milk Yield and Milk Composition Traits in Dairy Goats Using Machine Learning and Prior-Information Models
Simple Summary Improving milk yields and milk composition is a central objective of dairy goat breeding. This study compared conventional genomic prediction models, Bayesian regression models, and machine learning algorithms for three economically important traits in dairy goats: milk yield, milk fat percentage, and milk protein percentage. The results showed that the prediction accuracy depended jointly on the trait, genotyping strategy, statistical model, and type of biological prior information. Bayesian and gradient boosting methods were particularly useful for milk composition traits, whereas chip-based genotyping remained cost-effective for milk yield. These findings provide a practical framework for integrating machine learning, low-coverage whole-genome sequencing, SNP chip data, GWAS signals, and selection signature information into genomic selection programs for dairy goats.