Design of microarray experiments

Dr Penny Sanchez with G Glonek and A Metcalfe

Microarrays are a powerful technology that enables the measurement of the expression levels of many thousands of genes simultaneously. Rigorous statistical design of microarray experiments is critical for realizing the full potential of this technology.

Our methodology is aimed at the recommendation of optimal designs for a diverse range of two-colour microarray experiments. Based on statistical efficiency, the criterion of Pareto optimality is employed to optimise for the effects of particular scientific interest. A design is defined to be Pareto optimal if there is no other design that leads to equal or greater precision for each of the effects of scientific interest and greater precision for at least one such effect.

For elementary experiments, an exhaustive search for Pareto optimal designs can be carried out, as presented in Sanchez and Glonek (2009). However, this is infeasible for larger and more complex experiments.

To cater for situations for which an exhaustive search is infeasible, the methodology we have developed is an adaptation of the multiple objective metaheuristic method of Pareto simulated annealing to the microarray context. Based on a guided search, to explore the design space in an efficient way, our method is carried out to generate near-optimal designs in a relatively short amount of time. This is crucial to enable the recommendation of optimal designs that make the most effective use of resources for complex microarray experiments.

Investigators and affiliations

Dr Penny Sanchez, University of South Australia
Assoc Prof Gary Glonek, University of Adelaide
Assoc Prof Andrew Metcalfe, University of Adelaide

Publications

P. S. Sanchez; G. F. V. Glonek, Optimal designs for two-colour microarray experiments, Biostatistics 2009; doi:10.1093/biostatistics/kxp012.

Areas of study and research

+ Click to minimise