Comparison of prediction performance of LMM (blue), BVSR (red), and BSLMM (purple) in two simulation scenarios, where all causal SNPs are included in the data.

<p>Performance is measured by Relative Predictive Gain (RPG). True PVE = 0.6. Means and standard deviations (error bars) are based on 20 replicates. The x-axis show the number of causal SNPs (Scenario I) or the number of medium/small effect SNPs (Scenario II).