Skip to main content
Figure 1 | SpringerPlus

Figure 1

From: An omnibus permutation test on ensembles of two-locus analyses can detect pure epistasis and genetic heterogeneity in genome-wide association studies

Figure 1

Performance of MDR, RF and 2LOmb in the problem with 20 SNPs. The results are averaged over 30 independent simulations. MDR explores only models that do not contain more than 10 SNPs. The MDR output contains the most parsimonious SNP combination that yields the maximum prediction accuracy. The number of trees in RF is set to 100. The RF output consists of top-ranked SNPs, which are SNPs with variable importance in the top five percentiles of a normal distribution (Strobl et al. 2009). Association detection is declared for 2LOmb if the global p-value used as the detection indicator in its result is less than 0.05. The results from MDR, RF and 2LOmb are displayed using red diamond, blue triangle and black square markers, respectively. In each chart, the meeting point between two dotted lines denotes the graphical location representing ideal performance of the algorithm. Ideally, the algorithm should report only the causative SNPs in its output. In other words, both number of output SNPs and number of correctly-identified causative SNPs should be equal to the number of causative SNPs. The charts on which the red diamond markers are invisible denote the situations in which the performance of MDR and 2LOmb is similar.

Back to article page