DESeq2 interaction terms RIPseq analysis
Entering edit mode
sg197 • 0
Last seen 9 minutes ago
United Kingdom


I've found a few posts on this topic already (DESeq2 testing ratio of ratios (RIP-Seq, CLIP-Seq, ribosomal profiling), DESeq2 for Ribosomal profiling analysis [two-factor designs]) so I'm sorry that I'm probably asking the same question over. However I just want to be sure I'm understanding it correctly.

I'm analysing RIPseq data, and my model is this: ~batch + group + assay + group:assay. Groups = A,B,C,D and assay= Input or trap. My understanding is that the interaction term (group:assay) allows me to see if there are differences between the groups, accounting for the ratio in trap/inp.

From resultsNames(dds) I have:

 [1] "Intercept"                    "batch_2_vs_1"             "batch_3_vs_1"             "batch_4_vs_1"             "group_B_vs_A" 
 [6] "group_C_vs_A" "group_D_vs_A"  "assay_Trap_vs_Inp"            "groupB.assayTrap"      "groupC.assayTrap"    
[11] "groupD.assayTrap"

Firstly, would the below contrast give me the difference in expression between group B vs group C accounting for the ratio of trap/inp?

res <- results(dds, contrast=list("groupB.assayTrap", "groupC.assayTrap"))

Secondly, how would I achieve the same results for example comparing group A to group C using the resultsNames outputs? As groupA is the reference, is it just as below?

res <- results(dds, contrast=list("groupC.assayTrap"))

Thanks for any help or suggestions!

DESeq2 RIPSeq • 161 views
Entering edit mode
Last seen 20 minutes ago
United States

Yes that gives the comparison to reference.

Entering edit mode

Hi Michael, thanks for your reply. I've since looked into the extended examples in the help docs and I'm wondering whether the interaction model should be the other way around? So that the assay (Trap or Inp) is equivalent to "genotype" in the example doc, and the experimental group (cells treated under different conditions) is equivalent to "condition" in the example.

If I have my model as such:

~batch + assay + group + assay:group.

To determine if the group effect (treatment condition) is different across trap v input I'd use:

results(dds, name="assayTrap.groupB") ))

Or to test the same but for group BvC:

results(dds, contrast=list("assayTrap.groupB", "assayTrap.groupC"))

Also, is it possible to see the assay effect using the above model, for example the changes in gene expression for trap v inp specifically for groupA. Or for this, would it be simpler to merge group_assay and use a simple design model ~batch + group_assay?

Again thanks for any help!

Entering edit mode

I really have to defer to analysts for setting up the model. There are a lot of users and I have to limit myself to helping with the software. There are also multiple equivalent ways to set up tests and so it's important you work with someone locally so that you interpret the results properly.


Login before adding your answer.

Traffic: 605 users visited in the last hour
Help About
Access RSS

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6