Our estimates tend to gene, G6PD, we could not make any inferences, because we could not reproduce the observed pattern of diversity using simulations of positive directional selection.
http://plosgenetics.org/article/info:doi/10.1371/journal.pgen.1003011#article1.body1.sec3.sec2.p2

In the second sentence of the second paragraph of the section “Conclusions on data applications” of the Discussion, part of a sentence has been omitted. It should read: “Our estimates tend to be slightly older for several genes, which might be due to us assuming a recent bottleneck and thus a period of relaxed selection, whereas the majority of other authors assumed a constant population size for their timing inferences [49,75], which results in a stronger signal of selection (Figure S7). For one gene, G6PD, we could not make any inferences, because we could not reproduce the observed pattern of diversity using simulations of positive directional selection.”