Dear list: I am working on a small data set that came from cytokine membranes. Each membrane contains 23 specific cDNA fragments spotted in duplicates. The experiment design includes two treatment of interest, stimulant (pre-activated or non-activated cells) and inhibitor (w/ or w/o inhibitor). After manual normalization based on control spots on each membrane, I like to run limma as factorial design. My question is how I should handle the duplicate spots (block effect) in limma processing? Thanks! JP-