What does one-sided p-value in DESeq2 mean in biological terms?
1
0
Entering edit mode
nash12 • 0
@b501e59c
Last seen 9 months ago
United Kingdom

Dear All,

I am trying to understand what does one tailed p-values calculated in a DESeq2 analysis by setting the lfcThreshold to 0 and specifying the alternative hypothesis (as specified here: post) mean in biological terms. In a standard DESeq2 analysis we test for whether a gene is differentially expressed (DE) or not and a two-sided testing is done. Does the one-sided testing mean that we test whether a gene is DE and is up-regulated or DE and is down-regulated depending on the specification of the alternative hypothesis? Is my understanding correct or am I missing something here?

Also, in case I have multiple RNA-seq studies and I want to do a meta-analysis, can the one-sided p-value obtained from DESeq2 for a gene in each of the studies be used for combination of p-values using the Fisher or Stouffer's method? As is the case in microarray data where one can use one-sided p-values from individual studies for a gene. Or do I need to discard the genes with inconsistent expression direction across studies in case of RNA-seq meta-analysis for DE?

Thank you.

0
Entering edit mode
@mikelove
Last seen 1 day ago
United States

To Q1, yes.

What you do with the p-values from DESeq2 is really up to you. Here I focus on software questions.