Hi helpful people on the forum I'm having trouble to understand which is the correct way of designing the matrix for my 3factors (genotype, age, treatment) x2 levels experiment. i believe that all 3 factors are important and we are also interested in determining the interaction effects. according to the vignette; i defined the dds as recommended dds$group <- factor(paste0(dds$genotype, dds$treatment,dds$age)) design(dds) <- ~ group dds <- DESeq(dds) resultsNames(dds)

however i do note that i get different output from resultsNames(dds) if i did this

dds <- DESeqDataSet(gse,design= ~ genotype+treatment+age+genotype:treatment+genotype:age+genotype:age:treatment)

which i dont really understand.

I did a LRT to see if agegroup is an important factor; and my p-adj value is less than 0.05 for one of the gene. Does it mean i cannot reduce the model to exclude the both age as a main effect and interactions w age? be it a 2 (genotype:age) or 3 way (genotype:age:treatment) interaction?

designC <- as.formula(~ age)
ddsObjC <- DESeq(dds, test="LRT", reduced=designC)
LRT<-results(ddsObjC)

Questions about the best statistical design for your experiment are beyond the scope of what support I can provide on this site, I’d recommend working with a statistical collaborator to come up with the statistical analysis plan.