DESeq2 2x2 Design and Interaction Effect

Hi All,

I’ve been searching for answers to this for awhile now. DESeq2 in R seems (I think) to have the ability to input a 2x2 factorial design and retrieve differentially expressed genes significant by interaction. I can’t seem to figure out how to perform that in Galaxy though. If any has a tutorial or tips please let me know. Below is what I have tried so far.

My experiment is setup as follows:


I have tried defining the following:

Factor 1 as Healthy vs Disease with:
Healthy as Factor Level 1 and Disease as Factor Level 2

Factor 2 as Vehicle vs Drug with:
Vehicle as Factor Level 1 and Drug as Factor Level 2

When I run it in galaxy, I define all healthy samples (both healthy + vehicle and healthy + drug) as part of Factor 1 - Factor Level 1; all diseases samples (both disease + vehicle and disease + drug) as part of Factor 1 - Factor Level 2; all vehicle samples (both healthy + vehicle and disease + vehicle) as part of Factor 2 - Factor Level 1; and all vehicle sample (both healthy + drug and disease + drug) as part of Factor 2 - Factor Level 2.

I only ever get differentially expressed genes between the Healthy and Disease - none of the individual pairwise comparisons or the interaction term like you would in a 2-way ANOVA?

Does anyone have a method for obtaining this or a tutorial that I could refer to? Any help would be really appreciated!

Hi @Alex_Green

Limma and EdgeR both are by the same authors, and support more options (“contrast”). Examples are in the help sections on the tool forms. The end-to-end tutorials here have an example: