Fig. 1From: Integrating genomic and infrared spectral data improves the prediction of milk protein composition in dairy cattlePrediction R-squared for milk protein traits (TP: true protein nitrogen; TCN: total casein; TWP: total whey protein; \(\kappa \text {-}\)CN: \(\kappa \text {-}\)casein; \(\beta \text {-}\)CN: \(\beta \text {-}\)casein; \(\alpha _{S1}\text {-}\)CN: \(\alpha _{S1}\text {-}\)casein; \(\alpha _{S2}\text {-}\)CN: \(\alpha _{S2}\text {-}\)casein; \(\beta \text {-}\)LG: \(\beta \text {-}\)lactoglobulin; \(\alpha \text {-}\)LA: \(\alpha \text {-}\)lactalbumin) from multiple kernel learning using repeated random sub-sampling cross-validation. \({\mathbf {S}}\): spectral relationship matrix; \({\mathbf {G}}\): genomic relationship matrix; \({\mathbf {A}}\): numerator relationship matrix; DIM: days in milk; Top3SNP: top three markers with the largest effects.Back to article page