4
$\begingroup$

I'm running stageR as part of a pipeline based on DRIMSeq and wanted to know what it means when the transcript level padj is equal to 0, particularly when the gene is non-significant. I'm pretty sure that in some cases if the padj is 0 it means that the chance of it being different due to chance is 0. There are a lot more 0s than I would expect in sites where the gene padj is non-significant. One example, I have a gene padj that's >0.9 but both of the sites I'm looking at have a padj of 0.

I have a couple of suspicions about why this is: either because it defaults to this if the comparative difference is or it uses 0 as a substitute to NA. Am I thinking along the right lines? Are the 0s real or are they likely to be a quirk of the package?

$\endgroup$
4
$\begingroup$

In stageR, we automatically return stage-wise adjusted p-values of zero for transcripts coming from genes that were found to be significant and only have 1 transcript, when testing for differential transcript expression (DTE). Similarly, when testing for differential transcript usage (DTU), we return p-values of zero for transcripts derived from a significant gene consisting of no more than two transcripts (based on the tx2gene object you provide).

Transcript-level p-values for non-significant genes should not occur. We provide NA's for genes with only one transcript for DTU, and by default you should get NA p-values for non-significant genes, since these genes do not proceed further than the screening stage.

Could you provide some more details on the analysis? Which multiple testing correction have you selected?

Would it be possible to provide a reproducible example for your analysis?

$\endgroup$
3
  • $\begingroup$ I'm using DTU and I've set the alpha to 0.5 as we were hoping that some of the genes which were non-significant overall might have some features that were significant. I'll see if I can make a reproducible example in a bit. $\endgroup$
    – Sethzard
    Dec 5 '17 at 16:06
  • $\begingroup$ I'm really sorry it took so long to get back to you, I ended up getting distracted by something else. It turns out that the issue was probably elsewhere within what I was doing. $\endgroup$
    – Sethzard
    Dec 15 '17 at 13:36
  • $\begingroup$ No problem. Let us know if you have any remaining questions or issues. $\endgroup$
    – KoenVdB
    Dec 17 '17 at 20:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.