DESeq2 : results on combination of several levels of a factor

Question

[denis]

Dear DESeq2 experts,

I have RNA-Seq data replicated from different genotypes. Let's say for the sake of example

sample   genotype
samp1   A
samp2   A
samp3   B
samp4   B
samp5   C
samp6   C
samp7   D
samp8   D

and so on. I can easily compare each genotype against each other. Now let's say these genotypes have interesting common phenotypes (A and B have phenotype 1 and C and D have phenotype 2) and I want to compare combination of genotypes to find DEG that could explain the phenotypes. After reading DESeq2 vignette, manual, and several posts here and elsewhere, the only way I could think of to achieve this would be to write my condition file as :

sample   phenotype
samp1   1
samp2   1
samp3   1
samp4   1
samp5   2
samp6   2
samp7   2
samp8   2

then run the results function with contrast 2 versus 1. Can someone confirm this is the right and only way of doing it or am I missing something obvious here?

Answer 1

[swbarnes2]

I don't think that's the only way; I bet you can do something cute with lists in the contrast argument to results, but your way is probably the simplest.

[denis]

Thanks for your answer. I looked into this way and I used to do something like this by the past:

res = results(dds, contrast = list(c("genotypeD", "genotypeC"), c("genotypeB","genotypeA")))

Only problem is that now it does not work because resultsNames() returns "genotype_D_vs_A", "genotype_C_vs_A" and so on and that's the only names DESeq2 accept with a list as value for the contrast argument

[Carles Borreda]

Hi Denis

Have you managed to find a solution? I just ran in the exact same problem and I was wondering if you found a way to overcome this limitation

EDIT: Well minutes after writing this I found the following link where they give a working solution

https://support.bioconductor.org/p/101096/