Computational methods for the prediction of drug toxicity.
M Cronin · PubMed · 2000
Computational methods to predict drug toxicity are reviewed. Toxicological information can be retrieved from databases that enable the user to search for existing data. Quantitative structure-activity relationships (QSARs) provide predictive models for specific toxicity endpoints, membrane permeability and drug metabolism. Recent advances in the use and development of expert systems for toxicity prediction (eg, TOPKAT, CASE, DEREK and MetabolExpert) are reported. Approaches to predict receptor-mediated toxicities (such as estrogenic effects) are reviewed in the light of their being used as screens for combinatorial libraries. Overall strategies for, and future directions of, computational drug toxicity prediction are also suggested.