[Genetic algorithms in 3D-QSAR: Predicting the toxicity of dibenzodioxins, dibenzofurans and biphenyls]

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) and related compounds represent serious environmental health hazards, whose effects include dermal toxicity, immunotoxicity, tumor promotion, developmental and reproductive toxicity. There is strong evidence that the toxicity is mediated by the Aryl hydrocarbon (Ah) receptor, a regulatory element involved in the mammalian metabolism of xenobiotics. Using Quasi-atomistic receptor modeling, a 3D-QSAR technique, we have generated a receptor model for the Ah receptor. The model was trained using 76 dibenzodioxins, dibenzofurans, and biphenyls, and tested using 26 compounds different therefrom. 24 of these test compounds (92.3%) were predicted to within a factor of 10 of the experimental binding affinity. One compound was predicted false positive, another false negative (7.7% total). The sensitivity of the model with respect to the biological activity was demonstrated by means of a scramble test with negative outcome. Quasi-atomistic receptor modeling can be used for the testing of chemicals and pharmaca for a toxicological potential in an early development phase and thereby replace stressful tests on animals