Factorial and time course designs for cDNA microarray experiments
by
Patty Solomon
Centre for the Molecular Genetics of Development, University of Adelaide

Microarrays are a powerful new biotechnology for measuring the expression levels of many thousands of genes simultaneously and have important applications in the biological, agricultural and pharmaceutical sciences. There are myriad sources of uncertainty in microarray experiments and rigorous statistical design and analysis are essential for fully realizing the potential of these valuable resources. Early experience has shown that whilst classical experimental designs have much to offer this emerging multi-disciplinary area, new approaches to the design of microarray experiments are needed. Two of the questions most frequently asked by biologists on the brink of conducting a microarray experiment are Which mRNA samples should be competitively hybridised together on the same slide?' and How many times should each slide be replicated?' In this talk, I will describe joint work with Dr Gary Glonek which addresses these and related questions by proposing optimal designs for factorial and time course experiments. Our main criterion for optimality is statistical efficiency and our recommended designs are based on a new notion of admissibility for experimental designs. The approach enables appropriate designs to be selected subject to information on the contrasts of primary interest, the number of arrays available for experimentation, and other practical constraints. These new designs are superior to both the popular reference designs and to designs incorporating all possible multiple direct comparisons. Moreover, the proposed designs represent a substantial practical improvement over classical experimental designs which work in terms of standard interactions and main effects.

Date received: May 21, 2002