Deseq covariates with interaction

Question

[Folder40g]

Hello

I have a design with two factors, each one with two levels. And I want to know the effect of each variable and if the interaction effect.

treatment <- c("dox_no","dox_no","dox_yes","dox_yes","dox_no","dox_no","dox_yes","dox_yes","dox_no","dox_no","dox_yes","dox_yes")
aminoacid <- c("aa_no","aa_yes","aa_no","aa_yes","aa_no","aa_yes","aa_no","aa_yes","aa_no","aa_yes","aa_no","aa_yes")

comparison <- cbind(treatment, aminoacid)
rownames(comparison) <- c(paste("RNA", c(1:12),sep=""))
comparison <- as.data.frame(comparison)

comparison$treatment <- factor(x = comparison$treatment, levels = c("dox_yes","dox_no"))
comparison$aminoacid <- factor(x = comparison$aminoacid, levels = c("aa_yes","aa_no"))

Then I describe the model; RNAtable is a data.frame with 12 columns (samples) and 24015 rows (genes)

RNAcount <- DESeqDataSetFromMatrix(countData = RNAtable, colData = comparison, design = ~  treatment + aminoacid + treatment:aminoacid)

deseq <- DESeq(RNAcount)

resultsNames(des) 
"Intercept" "treatment_dox_no_vs_dox_yes" "aminoacid_aa_no_vs_aa_yes" "treatmentdox_no.aminoacidaa_no"

I believe that the effect for the aminoacidad and its differential expressed genes can be extracted such way

results(deseq, contrast = c("treatment","dox_no","dox_yes"))

Is this right? How can I get the effect for the other variable and for the interaction?

Thanks

Answer 1

[Devon Ryan]

results(deseq, name="aminoacid_aa_no_vs_aa_yes")

What you wrote would return the results due to treatment (you don't need the contrast, just specify the coefficient name).

Answer 2

[Folder40g]

So,

This results(deseq, name="aminoacid_aa_no_vs_aa_yes") would give the effect on each gene due to the effect a aminoacid.

This results(deseq, name="treatment_dox_no_vs_dox_yes") would give the effect on each gene due to the effect of the treatment.

Am I right?

I do not know hot to interpret this results(deseq, name=""treatmentdox_no.aminoacidaa_no"), I understand the meaning of intereaction but I do not know hot to interpret this comparison.

Thanks once more.

[Devon Ryan]

That's the additional change due to no doxycycline combined with no amino acid compared to what would happen if those two different conditions did not interact.

[Folder40g]

Thanks, I'll change the order of the factors so it will be easier to understand what the effect of the interaction means.