Answers are grounded in this paper and cite their sources.
SNP-weighted genomic relationship matrices (GRMs) may better reflect heterogeneous marker effects, but their performance depends on population and weighting intensity. We evaluated genome-wide association study (GWAS)-informed weighted genomic best linear unbiased prediction (GBLUP) in the red-brown line of Korean native chickens for 8-week body weight, 10-week carcass weight, and carcass yield. SNP effects were estimated using mixed linear model association (MLMA) and BayesC, and three weighting schemes were applied to construct weighted GRMs. Predictive performance was assessed using five repetitions of five-fold cross-validation, and weighting intensity was evaluated across 18 λ levels from unweighted to fully weighted models. Weighted models modestly increased predictive ability for body and carcass weight, whereas responses for carcass yield were small and inconsistent. Across traits, weighting generally increased the root mean square error (RMSE) and mean absolute error (MAE), reduced dispersion slopes, and increased reliability. Increasing λ altered estimates of additive genetic variance and heritability, particularly under squared-effect Bayesian weighting, and increased GRM diagonal heterogeneity. These results suggest that GWAS-informed SNP weighting did not consistently improve prediction in this population, and that its evaluation may benefit from considering prediction errors, calibration, and variance estimates alongside predictive ability.
No takes yet. Share an insight, caveat, or question.
Lee et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: