Surrogate variable analysis for paired RNA seq experiment

Question

[nhaus]

Hi all,

I have an RNAseq experiment with 1000s of samples. Each sample comes from a patient and was treated either with a drug or DMSO (as control). I am currently deciding between SVA and RUVseq to account for "hidden" technical variation. My question is, if I have identified surrogate variables (regardless of method) and include them in my model like this:

design = ~ X1 + X2 + X3 + X4 + X5 + X6 + X7 + ... + condition

Will the surrogate variables also account for the fact that I have a "paired" design (each sample has a control and a treated counterpart)?

Any insights are very much appreciated!

Cheers!