Hello, I am working on RNASEQ data with 4 treatments(Control, ABA, salt,sorbitol), 3 genotypes(col.0, lbd16-1, ds2.3) and three replicates. To test differently expressed genes of treated/untreated effect between lbd16 and col.0, I wrote my script like this:

dds <- DESeqDataSetFromMatrix(countdata, colData, design= ~ genotype + treatment + genotype:treatment)
dds$treatment <- relevel(dds$treatment , "Control")
dds$genotype <- relevel(dds$genotype , "Col.0")

keep <- rowSums(counts(dds)) >= 1
dds_F <- dds[keep,]

dds2 <- DESeq(dds_F)
dds2$group <- factor(paste0(dds2$genotype, dds2$treatment))
design(dds2) <- ~ group
resultsNames(dds2)

[1] "Intercept"                         "genotype_ds2.3_vs_Col.0"           "genotype_lbd16.1_vs_Col.0"        
 [4] "treatment_ABA_vs_Control"          "treatment_NaCl_vs_Control"         "treatment_sorbitol_vs_Control"    
 [7] "genotypeds2.3.treatmentABA"        "genotypelbd16.1.treatmentABA"      "genotypeds2.3.treatmentNaCl"      
[10] "genotypelbd16.1.treatmentNaCl"     "genotypeds2.3.treatmentsorbitol"   "genotypelbd16.1.treatmentsorbitol".

My question is:

1. Is it the right script to find DEGs among treaments and genotypes?
2. will res<-results(dds2, contrast=list( c("treatment_ABA_vs_Control","genotypeds2.3.treatmentABA"))) give me the result that differently expressed genes of ABAvscontrol between ds2.3 and col.0. If not, what would be the right one?
3. If I just want to see DEGs of ABAvscontrol of Col0, is this right? results(dds, contrast=c("treatment","ABA","Control"))
4. If I just want to see DEGs of ABAvscontrol of lbd16, is this right? results(dds, name="genotypelbd16.1.treatmentABA")

Take a look first at the vignette section on interactions.

2) for this you would use the interaction term genotypeds2.3.treatmentABA. This is the difference in treatment effect btwn ds2.3 and the reference genotype (col.0). The interaction terms give these differences in treatment effects that you are interested in.

3) In an interaction model with three genotypes, you have three effects of treatment. There isn't a general treatment effect, there is the effect in the reference genotype, and the other two.

4) To obtain the treatment effect in a non-reference genotype (not the difference among genotypes) you would use the list(c("...","...")) paradigm, adding the main effect and the interaction term. See the vignette section on interactions for a better idea on this.