Self-adaptive scouting-autonomous experimentation for systems biology

Abstract

We introduce a new algorithm for autonomous experimentation. This algorithm uses evolution to drive exploration during scientific discovery. Population size and mutation strength are self-adaptive. The only variables remaining to be set are the limits and maximum resolution of the parameters in the experiment. In practice, these are determined by instrumentation. Aside from conducting physical experiments, the algorithm is a valuable tool for investigating simulation models of biological systems. We illustrate the operation of the algorithm on a model of HIVimmune system interaction. Finally, the difference between scouting and optimization is discussed. © Springer-Verlag 2004.