I am analyzing 30 RNA-seq samples (6 donors in 5 treatment groups), and after creating an MDS plot, I noticed a clear donor effect. How does one deal with donor effect, if at all? Thanks in advance.

By blocking on it; either in the design matrix directly or, if your donors are confounded with your treatment, using duplicateCorrelation. Look at Section 18.2.9 of the limma user's guide for an example of the former (blocked on batches), and Section 17.3 for an example of the latter (blocked on donor).