Authors have taken into account most of the suggestions, and the article has been greatly improved.

To my opinion, a few minor points have to be improved before final acceptance: - Significant numbers should be only 2 or 3 instead of 4 in the AUC (with 4 numbers, it gives the false impression that AUC is an extremely precise measurement) - Each method has pros and cons. I think it is important to acknowledge also the limitations of the method, and eventually explain why and propose ideas for improvement. The fact that the method is the worst one when focusing only on top genes (FRP < 0.05 or in a lesser extent FPR < 0.1) should be highlighted and described. Maybe, this is not the best method when focusing on “top prediction”. All this should be described and commented in the main text (not only in the response to reviewers).