When we explicitly tested the potential confounding role of gene size, previously suggested [23], [25], we observed that large genes tended to receive more significant scores than small genes in the randomized data set (Figure S2).
