Fig. 3From: Efficient genomic prediction based on whole-genome sequence data using split-and-merge Bayesian variable selectionPrediction accuracies achieved using increasingly more variants in the merged dataset for SCS, PY and IFL. Merged variants were either included alone (blue dots) or combined with the 41,682 50k-chip SNPs (green triangles). Horizontal lines indicate the prediction accuracy obtained using the average GEBV across subsets (dashed black), GEBV computed using all 4,154,064 variants (dot-dashed grey), or the 50k SNPs (red dashed)Back to article page