Fig. 3
From: Genomic prediction with whole-genome sequence data in intensely selected pig lines
Effect of training set size on the genomic prediction accuracy for each set of preselected WGS variants in the three largest lines. The difference between each set of preselected WGS variants (Top40k, ChipPlusSign, and ChipPlusSign*) and the marker array (Chip) is shown. Red dashed line at ‘no difference’. Regression coefficient (b) and p-value of training set size is provided, as well as the coefficient of determination (R2) of the model. Results for traits ADG, BFT, and LD were jointly analysed, and the linear model included the trait effect (panels a to c) or the trait and line effects (panel d)