Nesting a covariate for differential expression design

Question

[RNAseqer]

Hello all,

I would like to create a design for differential gene expression analysis along the lines of

model.matrix(~0 + group + sex + batch)

but one where 'batch' is a nested variable that takes into account the library preparation batch, and the fact that some samples were re-run and the data combined to generate an adequate number of counts, AND the particular flow cells the one (or two) runs of each sample were performed on.

I have not used such a nested design before and though there is some good material I have found on the subject I am still unsure exactly how to encode such a complex nesting variable. I would very much appreciate any advice or examples that more experienced readers could provide.

Many thanks in advance!

[swbarnes2]

If you ran the exact same library more than once, just combine the counts. It's not a significant source of technical artifacts. Neither is flowcell.