How to adjust the p values of edgeR?

[wmseq]

Hi every one,

I am trying adjust the p values of my two-treatment comparison by edgeR. I am not sure if FDR need to be calculated with topTags before the adjustion, because exactTest just gives the p values without FDR.

In addition, how can I extract the DE genes with padjusted <= 0.05 from the following dataframe.

logFC logCPM PValue FDR padjusted
GB49890-RA 5.557823 7.1075888 1.018805e-43 1.978520e-40 1.978520e-40
GB46236-RA 4.667221 0.7197219 7.988596e-19 5.921318e-17 5.921318e-17
GB52184-RA 4.516693 1.3326365 1.519341e-23 2.269662e-21 2.269662e-21
GB50109-RA 4.477469 3.9928677 3.167978e-07 2.177004e-06 2.177004e-06
GB48922-RA -4.305247 2.7841370 4.546438e-29 2.006632e-26 2.006632e-26
GB40248-RA 4.127615 7.1749037 9.026131e-28 3.130133e-25 3.130133e-25

Thanks a lot!!

Richard

[dpryan]

topTags outputs adjusted p-values in the FDR column. The default method is BH, which I think is also the default for p.adjust().

For subsetting either type "help(subset)" or just google "R data frame subset" for other methods.

[wmseq]

Thank you very much, Devon!
I got it.

[wmseq]

Hi Devon,
After I finished my analysis, I got over 700 upregulated genes.
Do you think that the number is reasonable?
Thanks!

Richard

[dpryan]

Hard to say. Reasonable is pretty dependent on the underlying manipulation and sample number. If you're studying something like cancer (or anything else where you expect a pretty big number of differences) then that'd be entirely reasonable. I work on epigenetic inheritance, where that would be way too many DE genes.