Figure 5

Average ROC curves for different statistics when the number of affected pairs increases. Solid lines are computed from the Geninter rank p-values, dashed from the logrank test p-values. Results from simulations with different numbers of affected marker pairs are drawn with different colors. Each curve is an average (averaged over 20 repetitions) of an analysis of a cohort of 10,000 samples with the number of affected marker combinations varying between 1 and 20.