Question: Post-hoc filtering on the standard deviation in proportions
0
4 months ago by
Jack.Hearn0 wrote:

Hello,

I am following the "Swimming downstream" vignette for dtu using DRIMSeq and StageR and would like to clarify something. Does the output of section "5.3 Post-hoc filtering on the standard deviation in proportions" become the input for section "5.2 stageR following DRIMSeq" if performing the filtering? I think this is what is meant by "We do not recompute adjusted p-values, although we will provide the filtered p-values to the stageR procedure". With respect to section 5.2 of the vignette:

pConfirmation <- matrix(res.txp$pvalue, ncol=1) ​becomes pConfirmation <- matrix(res.txp.filt$pvalue, ncol=1)

Apologies for the minor point, but I want to be sure I'm understanding the vignette correctly.

Best wishes,

Jack

drimseq stager rnaseqdtu • 155 views
modified 4 months ago by Michael Love23k • written 4 months ago by Jack.Hearn0
Answer: Post-hoc filtering on the standard deviation in proportions
0
4 months ago by
Michael Love23k
United States
Michael Love23k wrote:

The first "rnaseqDTU" post!

Yes, you should put res.txp.filt\$pvalue as the single column of pConfirmation, for performing stageR adjustment on the filtered transcript results.

If you want to see more of the original code it is here (with the relevant chunk highlighted):

https://github.com/mikelove/swimdown/blob/master/dtu/dtu_analysis.R#L75-L88

Great, thanks for the swift reply! As a follow up I'm interested in contrasts from DRIMSeq so I assume the filtering and StageR steps need to be run for each contrast, that being the results of each "dmTest" performed.