Fig. 3

Comparison of prediction accuracy performances of GBLUP, mssBLUP, and wmssBLUP. For wmssBLUP, the \(w\) was 0.1, 0.6, 0.5, and 0.4, respectively. The prediction accuracy performance of each method was measured by the average Pearson correlation between predicted values and phenotypic values of 5 replicate in the validation subset. In each replicate, the dataset was randomly split into a reference subset containing 80% of individuals and a validation subset containing the remaining 20%. For each violin plot, the middle line represents the median value, and the upper and lower ends of each box represent the maximum and minimum