I'm running DESeq2 for my expression analysis and I'm having trouble with the design formula. I have 32 mice samples and 3 different factors and each factor has 2 levels. So the last factor (tissue) has always 4 biological replicates. Here is an overview:
What I'd like to compare is the following interactions: treatment:type + treatment:type:tissue. In special I'm interested in treatmentA.typechanged vs treatmentB.typechanged, treatmentAtypecontrol vs treatmentBtypecontrol, treatmentBtypechanged.tissuelayer_one vs treatmentBtypechanged.tissuelayer_two, treatmentBtypecontrol.tissuelayer_one vs treatmentBtypecontrol.tissuelayer_two
I run the following DESeq:
dds <- DESeqDataSetFromMatrix(countData = Jcounts, colData = coldata, design = ~ treatment + treatment:type + treatment:type:tissue)
dds <- DESeq(dds, parallel=T)
As a result I get
 "Intercept" "treatment_A_vs_B" "treatmentA.typecontrol" "treatmentB.typecontrol"
 "treatmentA.typecontrol.tissuelayer_one" "treatmentB.typecontrol.tissuelayer_one" "treatmentA.typecontrolchanged.tissuelayer_two" "treatmentB.typecontrolchanged.tissuelayer_two"
I was wondering if this is the right way to go? However resultsNames doesn't show all effects. Am I doing something wrong?
Another thing I could do is to run DESeq several times but exclude samples. For example for the first comparison I would run the sample from treatmentA.typecontrol.tissuelayer_one vs treatmentA.typecontrol.tissuelayer_two etc. but I would prefer to have all samples in one DESeq run.
In another post I read that it is also possible to combine two factors (merging the columns) which creates one single factor and run DESeq with that. In my case it would be a factor with the following levels treatmentA-typecontrol, treatmentA-typechanged, treatmentB-typecontrol, treatmentB-typechanged. Would this be possible?
Thanks for the help!