Fig. 15

Graphical illustration how the Bonferroni and the Benjamini–Hochberg correction work, for an example with 7129 tests and 0.05 as desired significance level in each case. Applying Bonferroni, only the results of the tests with p-values smaller than 0.05 / 7129 (represented by a dotted line) provide evidence against the null hypothesis. For Benjamini-Hochberg, the significant genes are those whose tests yield p-values smaller than the largest p-value under the threshold, circled in green in the figure. The threshold is represented by the dashed line. The line has intercept 0 and slope 0.05 / 7129, where now 0.05 is the desired level of FDR control