Conference on Applied Statistics in Agriculture

DESIGNING MICROARRAYS

Microarrays allow the simultaneous assessment of expression levels for thousands of genes across various treatment conditions and time. It has been shown that in these experiments expression levels can also be affected by factors in the printing of the slide, in the hybridization process, and in the post-hybridization process. Recently, variations of the incomplete block design were proposed as a way to avoid confounding the expression levels of interest with several of these nuisance factors. In this paper, we propose additional design procedures to remove factors that contribute to the spatial variability on a slide. This approach requires the use of replication, and involves designing how the replicates are to be positioned on the slide. We demonstrate these techniques using an experiment involving sixty-four genes, four replicates per slide, and five treatment conditions.