After going crazy on Bioconductor and edgeR guide, I thought to run it by you.

I am trying to analyze a data set. It is 7 cell-lines each cell line has a control and a treatment. So it's paired. I am interested in + and - differences only.

Cell A +
Cell B +
Cell C +
Cell D +
Cell E +
Cell F +
Cell G +
Cell A -
Cell B -
Cell C -
Cell D -
Cell E -
Cell F -
Cell G -

I was wondering what would be the correct way of doing the design matrix with model.matrix and subsequently the contrast matrix.

cell.line <- rep(LETTERS[1:7], 2)
treatment <- rep(c("Treat", "Con"), each=7)
design <- model.matrix(~cell.line + treatment)
# ... after voom and lmFit()...
topTable(fit, coef=ncol(design))

... assuming that the treatment effect is the same in all cell types.

No need to go crazy. Just follow Aaron's answer or else Section 9.4.1 of the User's Guide.