Fig. 4From: Genomic prediction based on selective linkage disequilibrium pruning of low-coverage whole-genome sequence variants in a pure Duroc populationHeat map of genomic prediction accuracy for three real traits using SNPs after SLDP. Heat map of genomic prediction accuracy for three real traits using SNPs after SLDP according to the r2 (from 1 to 0.05) and P-value (from 0.0001 to 0.01) gradient in the training population. AGE, age to 100Â kg live weight; BF, back fat thickness; TTN, total teat number. Panels a, c, and e show the results of the GBLUP model; panels b, d, and f show the results for the BayesR model. Each square represents a parameter combination and the accuracies are indicated with deeper colors corresponding to high (red) or low (blue) accuracy. Black dots mark the optimal combination of parameters for accuracy. The accuracy was obtained by five-fold cross-validation in the training population and defined as the Pearson correlation between GEBV and phenotypesBack to article page