The Particle Swarm Interval Rule Optimizer with an Application to Drug Design Data

Life sciences have become a wide area for applications of informatics. One prominent topic in life sciences is the design of new drugs. Chemists use combinatorial libraries for creating new molecules in laboratory, but the challenge is to find appropriate design proposals by computational means before spending costs and time in laboratory. Our aim is to generate rules for required molecular properties of specific classes of molecules. Since the encoded molecular data are high dimensional and often have complex structures, computationally intelligent methods constitute a coherent approach to the problem. In our contribution we propose a new optimization method for interval rules, the particle swarm interval rule optimizer (PSO-lntRule). Its characteristics are described and compared to the previously developed interval rule optimizers.