Combining P Values And Adjust For Multiple Hypotheses Or The Other Way Around?

Question

[lwc628]

I am trying to combine p values from the multiple expression data using fisher's method. The question is this:

Should I combine pvalues first and then adjust for multiple hypothesis testing? or use the adjusted p values and then combine them?

Answer 1

[brentp]

combine, then adjust.

Fisher's method has certain assumptions about what the p-value is that will almost certainly be violated if you do multiple-testing correction first.