Figure 6

Discoveries with Global FDR correction for Mann–Whitney U test. Mann–Whitney U tests with FDR correction were performed on unclustered (A, C, E) and clustered (B, D, F) datasets. In each panel, the number of discoveries with p < 0.05 is plotted in red crosses, while the average number of discoveries of the data permuted 100 times is plotted with open green circles. A) original dataset of 1350 peaks; B) original dataset after technical clustering at threshold 0.5; C) simulated dataset of each peak randomized within groups; D) simulated dataset from C after technical clustering at threshold 0.5; E) simulated dataset of each peak fully randomized; F) simulated dataset from E after technical clustering at threshold 0.5.