This is a test version of Biostars. For the public version, visit https://www.biostars.org.
p-value distribution of edger or deseq2

I have two datasets and did differential analysis and I am concerned about the distribution of the p-values.

The first dataset showed quite a weird p-values distribution. It seems to be uniform, meaning that all my hypothesis are null. What does it say about my data? Was there some kind of an uncultured batch effect or another confounding variable?

< image not found >

The second one seems to be ok:

< image not found >

One would also typically see a peak around 1 but why would there be a peak around 1?

rna-seq
• 3,177 views • link
updated by Ram Ram • written by tonja.r tonja.r

1 answer

  1. There might have been a batch effect, though it doesn't appear to be terribly large (they tend to lead to a fewer p<0.05 genes than you'd expect if nothing is DE). Having said that, I suspect that (A) there really are no changes or (B) you lack the power given the actual effect size to find them.
  2. One hopes to see a peak around 0, which is what I suspect you meant to say. You tend to see enrichment toward 1 when there's a batch effect.
• 0 views • link
written by Devon Ryan Devon Ryan

Why would one interpret enrichment toward 1 as a batch effect?

• 0 views • link
written by tonja.r tonja.r

Because that's a common cause.

• 0 views • link
written by Devon Ryan Devon Ryan

Log in to answer this question.

More posts like this