Additional file 1 of Integrating multi-layer perceptron and random forest in an ensemble framework for improved genomic prediction accuracy and SHAP-derived interpretability of residual feed intake in cattle

作者：Edwin Ong Jun Kiat, Mark Mooney, Faisal I. Rezwan, Hui Wang, Masoud Shirali · 发表于：Figshare · 年份：2026 · DOI：10.6084/m9.figshare.33188306 · 研究领域：Genetic and phenotypic traits in livestock、Effects of Environmental Stressors on Livestock、Genetic Mapping and Diversity in Plants and Animals

Additional file 1: Table S1. Fixed effects of covariates on total dry matter intake (kg DM/Wk) from the linear mixed effects model used to derive residual feed intake (RFI). Table S2. Summary of 190 SNPs associated with residual feed intake (RFI) using the Random Forest model. Table S3. Summary of 473 SNPs associated with residual feed intake (RFI) using the Multi-Layer Perceptron model. Fig. S1. Feature importance of top-ranked SNPs ordered by their mean absolute SHapley Additive exPlanations (SHAP) values derived from the Random Forest model. Blue bars indicate markers associated with a decrease in the estimated breeding value (EBV) for residual feed intake (positive effect), while pink bars indicate markers associated with an increase in EBV (negative effect). Fig. S2. STRING interaction network for genes linked to SNPs associated using Random Forest to residual feed intake in dairy cattle with the highest degree of centrality represented by node FTSJ3. Fig. S3. Feature importance of top-ranked SNPs ordered by their mean absolute SHapley Additive exPlanations (SHAP) values derived from the Multi-Layer Perceptron model. Blue bars indicate markers associated with a decrease in the estimated breeding value (EBV) for residual feed intake (positive effect), while pink bars indicate markers associated with an increase in EBV (negative effect). Fig. S4. STRING interaction network for genes linked to SNPs associated using Multi-Layer Perceptron to residual feed intake in dairy cat...