Application of TOPKAT and TEST software in prediction of chemical toxicity

Objective The predictability and applied range of TOPKAT and TEST software in the prediction of health and ecological toxicity of compounds were discussed, in order to easily, effectively and accurately use these softwares in the practice.

Methods The simplified molecular input line entry specification(SMILES)method, combined with 2 toxicity prediction software (TOPKAT and TEST) was employed to achieve the toxicity prediction of compounds. The predicted toxicity of selected chemicals was compared with the toxicity data published by European Chemicals Agency and the advantage and disadvantage of 2 softwares were present.

Results Both TOPKAT and TEST softwares could predict the healthy and ecological toxicity of most common compounds. It seemed the TEST software is prior to the TOPKAT software. TEST software could be used in the prediction of the toxicity of multi-substituted benzenes, heterocycles, and macromolecules, whiles TOPKAT software couldn't be used to predict the toxicity of pyridines, some phenols, and anilines. The accuracy of chemical toxicity classification based on predicted toxicity with these two softwares was approximately 85%. The TOPKAT software was prior to the TEST software in the prediction of toxicity of pesticides; the linear decision coefficient (R²) between predicted value of pesticides with TOPKAT software and the reported data was 0.852(health toxicity)(P < 0.01)and 0.826(ecological toxicity)(P < 0.01). The TEST software was prior to TOPKAT software in the prediction of toxicity of benzenes compounds; the linear decision coefficient (R²) between predicted value of benzenes with TEST software and the reported data was 0.932(health toxicity)(P < 0.01)and 0.787(ecological toxicity)(P < 0.01). Both TOPKAT and TEST softwares showed good predictability for the toxicity of esters and alcohols; the linear decision coefficients (R²) between predicted value of esters and alcohols and reported data were 0.854-0.986(health toxicity)(P < 0.01)and 0.821-0.981(ecological toxicity)(P < 0.01).

Conclusion Both TOPKAT and TEST are good at predicting healthy and ecological toxicity of most compounds effectively and accurately, which can provide the basis of toxicity assessment of new compounds and decision of drug development at early stage.