Two-Stage Machine Learning-Based GWAS for Wool Traits in Central Anatolian Merino Sheep

The study utilized a two-stage machine learning-based GWAS framework to explore the genetic basis of wool traits in 228 Central Anatolian Merino sheep. The first stage involved feature selection using methods like LASSO, Ridge Regression, and Elastic Net to create a consensus SNP panel for each trait. The second stage applied Random Forest and Support Vector Regression for association modeling, achieving predictive models with R² values up to 0.86. Key genes identified include MTHFD2L and EPGN for fiber diameter, COL5A2 and COL3A1 for staple length, and FAP and DPP4 for greasy fleece yield. The study highlights the effectiveness of machine learning-enhanced GWAS in identifying significant genetic markers, offering new targets for improving wool quality and yield in sheep breeding programs.