Genetic algorithms in 3D-QSAR: The use of multiple ligand orientations forimproved predictions toxicity.

In a recent article in this journal, we discussed the application of a 3D-Q SAR technique to the prediction of the toxicity of dibenzodioxins, dibenzof urans, and biphenyls (Vedani et al., 1999a). The use of such methods is leg itimate because there is strong evidence that the toxicity is mediated by t he Aryl hydrocarbon (Ah) receptor; a regulatory element involved in the mam malian metabolism of xenobiotics. In an extention to a concept developed at our laboratory (Vedani et al, 1998), ute now show that safer predictions a re possible if instead of a single - and therefore, necessarily biased - as sumption about the mutual orientation of the toxins, an ensemble of possibl e orientations is used for model construction. The contribution of a single entity within this ensemble to the toxin-receptor interaction energy is de termined by a Boltzmann criterion. While in the single-orientation model tw o of the 26 toxins of an external test set were predicted false positive or false negative, all test substances are correctly predicted in the multipl e-orientation model - including lip to four different orientations per mole cule - within a factor 10 of the experimental binding affinity. These results demonstrate that 3D-QSAR techniques based on a genetic algori thm can be used to predict the toxicity of chemical and pharmacological sub stances "in compute" if a receptor-mediated mechanism can be assumed. Conse quently, the method can be used in toxicological screening assays, thereby replacing stressful tests on animals.