pvalue vs adjust p-value

Question

[pingu77]

Hello,

I have a very naive question: I am running a gene expression analysis using DESeq2 and I plot the histogram of the p-values.

The histogram looks like the one in the scenario A of this blog http://varianceexplained.org/statistics/interpreting-pvalue-histogram/ (sorry I don't why but I cannot upload the jpeg of my plot).

However if I plot the histogram of my adjust p-values, it looks like scenario D, with several peaks between 0.60 and 1.

How is it possible? what does it mean from a statistical point of view?

Thank you in advance

Answer 1

[Nicolas Rosewick]

p-value histogram is only usefull for nominal p-value as p-value follows a uniform distribution under the Null hypothesis ( not the case anymore for adjusted p-value). So if you have a conservative histogram like figure A using your nominal p-value you should be ok to go further.