Alan Dabney and John Storey have pointed out that problems with FDR estimates in the Choe et al. paper stem from earlier problems with p-value distributions. This may explain why the FDR estimates were too low. They also argue that the distribution issues are caused by a flawed experimental design. See their comments and why we disagree with the latter point here. Read a more detailed treatment of our new analysis of normalization issues here and here. Fodor et al. show that non-uniform p-value distributions may be a common issue in microarray analysis and suggest some fixes in this Genome Biology paper.