For meaningful comparisons of multiple microarray experiments the initial overall design of the experiments is crucial. This includes the necessary controls, replicates etc. Even simple questions may demand 8 to 16 microarray experiments performed from individual biological samples that were processed in parallel. Some examples from the reference cited below are shown here. An example of a typical large microarray study that would be performed for researchers of the SFB is shown here. Data from an original experiment are shown here: pdf (68kB), xls (104kB).
Yang YH, Speed T. Design issues for cDNA microarray experiments. Nat Rev Genet. 2002 Aug;3(8):579-88