I am using the limma package for differential expression analysis. In my datasets there are strong confounding effects due to both technical variation and biological contamination of the tumor cells with infiltrating cells and stroma.
I would like to cluster my patient samples after removing the confounders and for this I would need a matrix that has the new values after removing the confounding effects (similar to what ComBat is doing).
However, I have 10 covariate terms and 15 of their pairwise interactions and Combat doesn't allow for more than one Batch effect.
Could I use limma to fit a linear model including all covariates and then calculate a new matrix that would give the gene values after removing the confounders, Or is there some way to do this using sva?
I would really appreciate your reply, I have been scratching my head for a couple of days now.