Question: DeSeq2 with three groups which approach to use? LRT or just do multiple pairwise comparisons?
gravatar for casey.rimland
10 months ago by
University of Cambridge, National Institutes of Health, Chapel Hill School of Medicine
casey.rimland100 wrote:

I have a question about the best approach for analyzing an RNA-Seq experiment using DeSeq2. My experiment involves three different tissue types (named C, G, and P). I want to find out what genes are enriched in/unique to  each tissue in comparison to the other two tissues.

What I was not sure of is which test in DeSeq2 is most appropriate to use: pairwise comparisons between each tissue or using the likelihood ratio option. 

Right now I have just been doing pairwise comparisons of:

C vs G,

C vs P,

P vs G.

Then I took the DEGs and looked for overlap in the gene lists using Venn Diagrams in order to get genes specific for each tissue type... For example:

100 genes are up-regulated in tissue type C in comparison to tissue G.

200 genes are up-regulated in tissue type C in comparison to tissue P.

Of these 300 genes, 75 are in common across both lists and therefore are potential "Tissue C specific" genes.

Is this the best way to go about this type of analysis? Or would using the Likelihood ratio test in DeSeq be a better option? My confusion with the LRT test is how do you go about finding genes that are up/down-regulated in each particular group compared to the others? 

Essentially how do I get at asking what genes are up-regulated in Tissue C vs. Tissue G and Tissue P... And so on for each of the other tissues...

I read through the vignette section for LRT tests and also looked on bioconductor previous posts. I found some posts with similar questions but after reading through them I still was confused on the best approach and would love any insight someone more experienced may be able to offer! Thanks :D

ADD COMMENTlink modified 10 months ago by Michael Love20k • written 10 months ago by casey.rimland100
gravatar for Michael Love
10 months ago by
Michael Love20k
United States
Michael Love20k wrote:

Performing pairwise Wald tests and looking at the intersection is my preferred way of specifying that A is different than B and A is different than C. 

ADD COMMENTlink written 10 months ago by Michael Love20k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 421 users visited in the last hour