I have the following design where each sample is pooled from many samples and the control and treated(S) samples are not paired. I want genes that are differentially expressed under the effect of the 2 drugs for treatment S. I created 'sample.nested' to account for the pairing of control and treated samples. I used a design
design(dds) = ~ drug + drug:sample.nested + drug:treatment
- Firstly, is my design right for the question I mention?
- Secondly, resultsNames(dds) gives a term "drugisovs_pro". Will this tell me the differentially expressed genes for drug 'iso' compared to 'pro'? Does this account for sample differences or treatment effects while talking about this difference? I use relevel in the design for dds$treatment and dds$drug, so that the reference levels are 'control' and 'pro' respectively.
- Thirdly, one of the questions I am interested to know is the drug specific effects. I guess in this above design I am controlling for the drug specific effects with ~drug. How can I do this. Please advice
sample treatment drug sample.nested U46801 S iso 1 U46813 control iso 1 U46802 S iso 2 U46814 control iso 2 U46803 S iso 3 U46815 control iso 3 U46804 S iso 4 U46816 control iso 4 U46917 S pro 1 U46929 control pro 1 U46918 S pro 2 U46930 control pro 2 U46919 S pro 3 U46931 control pro 3 U46920 S pro 4 U46932 control pro 4