I am trying to analyse a polysome profiling experiment using DESeq2 and I got stuck deciding on which design would be best to use.
The aims of my experiment are:
1- to investigate the effect of a drug (L-leucine) on translation
2-to look at mRNA translation in patients vs disease samples
I want to test for the ratio of the ratios using a LRT and two interaction terms and these are the groups I want to test:
type: total RNA (RNA), Ribosome-bound RNA (RBR)
treatment: control (D-leucine), treated (L-leucine)
disease: Healthy, 5q syndrome
I worked out a design based on previous posts and DESeq2 vignettes but I am not sure whether it's correct and whether I am extracting the results right.
Here it is:
Design(dds) < - ~type + treatment + disease + type:treatment + disease:treatment
dds <- DESeq(dds, test="LRT", reduced= ~ type + treatment + disease)
To get effect of treatment on ratio between RBR/RNA I do:
Then to find the interaction effect of condition:treatment effect across disease:
Results(dds, contrast=c(0 ,0 ,0 ,0 ,1 ,0))
This was based on this previous post https://support.bioconductor.org/p/76966/ but I am wondering:
1- to get the disease effect on the ratio between RBR/RNA should I maybe have this instead?
Design(dds) <- ~condition + treatment + disease + condition:treatment + condition:disease
Perhaps it's simply a different thing than disease:treatment but I am not sure which one to use in this case.
2- to get the disease effect on the ratio between RBR/RNA I used Results(dds, contrast=c(0 ,0 ,0 ,0 ,1 ,0)) but, since I cannot assume that the treatment effect on type is the same across disease, should I have
Results(dds, contrast=c(0 ,1 ,0 ,0 ,0 ,.5)) instead?
It is a quite complex design so any help would be very much appreciated.
All the best,