Users may sometimes find that their gamma adjusted p-values are NaN. This is because the permutation p-values could not be modelled properly by Gamma distribution. The issue arises usually because the number of permutations with significance is too close to zero.