Fig. 3

Testing and validation of prediction accuracies of genomic estimated breeding values (GEBVs) in fresh-market tomato yields. a Prediction accuracies of four different traits (Y, the total yield; XY, the yield of extra-large-sized fruit; SY, the yield of any fruit smaller than medium size; RY, the yield of red-colored fruit regardless of size) in 68 core inbred contemporary fresh-market tomatoes (inbred tomato set) are calculated using four different sets of SNPs and using an SVM-linear prediction model. Numbers 1 through 4 below each X-axis indicate four different sets of SNPs: 1, mapped SNP set (558 SNP sites); 2, regularization SNP set (364 SNP sites); 3, mapped/regularization SNP set (866 SNP sites); and 4, distributed SNP set (7266 SNP sites) (a and b). Colored rectangles indicate the 25th and 75th percentiles of the means of prediction accuracies; error bars indicate the ranges for the mean values (a and b). Prediction accuracies calculated using other models (rrBLUP, SVM-radial, SVM-poly, and random forest) can be found in Additional file 11. b Validation for prediction of Y, XY, SY, and RY traits using 420 F₂ individual plants derived from four different F₂ populations (F₂ population set) and using SVM-linear. Plots are color-coded according to figure a