Planning a scientific experiment generally requires balancing multiple objectives, and microarrays are no exception. A good microarray design will make efficient use of expensive resources. Design constraints may include absolute limits on the number of arrays or RNA quantities. Other important considerations are robustness to various kinds of missing data.
Microarray studies generally must include replicates to allow the assessment of biological variability and allow inference to the population of interest. Often there is a multi-factorial structure to the RNA samples. These ideas will be discussed and illustrated through case examples.