Fig. 5

q-q plot of p-values of SNPs. The blue curve is the p-values of candidates, calculated by McNemar’s test. The three linear curves show the significance level adjusted by the 0.05 false discovery rate, each based on different assumptions about the number of true null hypotheses. The yellow line assumes that the number of hypotheses is equal to the total number of SNPs in input data, which gives the most conservative possible adjustment. The orange curve assumes that the number of hypotheses is equal to the number of candidates produced by Feature Filtering and used for the statistical test, which gives the least conservative adjustment. The purple line estimates the number of hypothesis according to [36], a method for the estimation of the effective number of hypotheses, which is almost perfectly aligned to the less conservative approach