DESeq2 Factor Level Design

deseq2
#1

I have 3 muscles I am trying to perform differential expression with.

For each muscle type, I have 3 replicates.

How can I get pairwise comparisons between muscle 1 vs muscle 2, muscle 1 vs muscle 3, muscle 2 vs muscle 3. Within the DESeq2 tool I have factor set to muscle, then level 1 as muscle 1, level 2 and muscle 2, etc. Then within each level/,muscle type, I select the three replicate data count files.

I read that DESeq2 takes the first input as the reference (so that would be muscle 1) and then compares everything to that. But that’s not necessarily what I want. I have been using Cuffdiff to perform analyses but wanted to use a more updated tool but I can’t seem to set it up as it needs to be. Can anyone give me some guidance?

2 Likes

#2

Trying to work through similar problems…

I’m aware of the math to an extent, but ultimately, when you’re building a model of expression change in a gene and comparing it to a normal distribution, the model incorporates each sample’s expression level, so idk how to get pairwise comparisons of LFCs/significance from a matrix with multiple factors/variables and the levels therein.

I have a bunch of RNAseq data from mutant mice at different ages; and the data clusters better after including age as an additional factor (genotype is my primary factor), and the test results are different if I remember right, but ultimately, how do you do a pairwise test (contrast) or examine interactions on Galaxy? The DESeq2 paper is very intuitive until you dig deeper into the methods. The documentation in Bioconductor for the DESeq2 R package is good, but is there the appropriate functionality here on Galaxy for this?

2 Likes

#3

It almost seems that it’s not available as part of the Galaxy wrapper. Ultimately, I ended up ditching DESeq2 and ended up using edgeR as I found that their set up made more sense to me. I also found it easier to manipulate.

2 Likes

#4

Yes, EdgeR and Limma offer more functions and the tool forms are structured a bit differently.

Search with the keywork “edger” here to review prior Q&A and tutorial links, including tutorials that are still in progress for final publication to the Galaxy Training Network (GTN): https://galaxyproject.github.io/training-material/

0 Likes