On 2014 Jan 23, Farrel Buchinsky commented:

Interesting article with interesting exploration. I noted that the determination of whether a gene was consistently upregulated or downregulated was done by the Mann-Whitney U test. At least 45 separate (at some point I read of 84 separate genes) Mann-Whitney U tests were done. To correct for multiple testing would it not be more robust to perform the Bonferroni correction in which the alpha threshold would be set at 0.0011? Pitty about the non-normality of the data, since in a Mann-Whitney (Wilcox) a slightly elevated result counts the same as a massively elevated result. Modeling the data with Bayesian statistics or doing some more replicates may settle it.

On 2014 Jan 23, Farrel Buchinsky commented:

Interesting article with interesting exploration. I noted that the determination of whether a gene was consistently upregulated or downregulated was done by the Mann-Whitney U test. At least 45 separate (at some point I read of 84 separate genes) Mann-Whitney U tests were done. To correct for multiple testing would it not be more robust to perform the Bonferroni correction in which the alpha threshold would be set at 0.0011? Pitty about the non-normality of the data, since in a Mann-Whitney (Wilcox) a slightly elevated result counts the same as a massively elevated result. Modeling the data with Bayesian statistics or doing some more replicates may settle it.