Figure 9From: Detecting purely epistatic multi-locus interactions by an omnibus permutation test on ensembles of two-locus analysesPrediction accuracy from the MDR analysis. A 10-fold cross-validation strategy is applied during the accuracy evaluation. The best MDR model is located by exploring all possible SNP combinations. All erroneous SNPs, which are left over after the screening by 2LOmb, have been successfully identified. All MDR models contain the correct number of causative SNPs. In addition, the MDR cross-validation consistency is 10/10.Back to article page