On the A-optimality criterion for finding two-color microarray optimal designs.

Several studies on two-color microarray experimental designs consider the A-optimality criterion for the assessment and/or search of appropriate/optimal designs. What seems to be widely ignored, however, is the well-known fact that A-optimal designs depend on the choice of scale or code combination of the X-variables. In this article, it is shown that intrinsically different optimal microarray designs will be obtained depending on the chosen code combination of the qualitative factors. Moreover, a substantial loss in efficiency will result when a design, which is optimal for a specific code combination, is used for another one. This coding problematic is also applicable to the admissibility concept.