Derivation of quantitative structure-toxicity relationships for ecotoxicological effects of organic chemicals: evolving neural networks and evolving rules

Organic compounds can produce toxicological effects on aquatic organisms through a variety of mechanisms of action (MOAs). These MOAs can be grouped into two major types: narcotic and reactive. A means to predict MOAs from organic compound structural parameters is important in developing an understanding of environmental toxicity and organism response. This paper presents results of using evolved neural networks or evolved rule-based classifiers to discriminate between narcotic and reactive MOAs of small molecules. The results indicate that both evolutionary methods are useful in this regard