I am trying to use DEXSeq::testForDEU() to compare two models with likelihood ratio test. The null model can be written as Y = β*0 + β*1 S + β*2 T + ε and the full model can be written as Y = β*0 + β*1 S + β*2 T + β_3 (S*T) + ε. In DEXSeq manual, it says "It must specify an interaction term between a variable from the sampleData columns with the ’exon’ variable". How can I add another interaction term between two variables to the full model? Current, I wrote the "fullModel" as "~sample + exon + S:exon + T:exon + S:T:exon" and "reducedModel" as "~sample + exon + S:exon + T:exon". Does it make sense to write full model's interaction term "S:T:exon" this way? Thank you.

Hi zhenzuo2. Definitely, you can test the hypothesis that you wrote down using exactly the formulae that you mentioned.