论文部分内容阅读
[目的]归纳TOPKAT、TEST两款毒性预测软件对有毒化学物健康和生态毒性的可预测性和适用范围,探讨高效、简便、快速的毒物检测实用方法。[方法]采用简化分子线性输入规范(SMILES),结合两款毒性预测软件对化学物的毒理学进行预测,并将预测结果与欧洲化学品管理署公布的毒理学数据进行对比,总结两款软件的优缺点。[结果]TOPKAT和TEST软件能预测大多数的常见化学物毒性,TEST软件的化学物预测覆盖率高于TOPKAT软件,能预测多取代苯环类、杂环类和大分子化学物,而TOPKAT无法预测吡啶类、部分苯酚类和苯胺类化学物的毒性。两款毒性预测软件在预测化学物是否具有毒性方面的正确率均在85%左右。TOPKAT软件对于农药类化学物的毒性预测正确率高于TEST软件,TOPKAT预测值与文献值的线性决定系数R~2为0.852(健康毒性)(P<0.01)、0.826(生态毒性)(P<0.01)。而TEST软件对于苯系化学物的毒性预测结果略优于TOPKAT软件,TEST预测值与文献值的线性决定系数R~2为0.932(P<0.01)(健康毒性)、0.787(生态毒性)(P<0.01)。对于酯类和醇类化学物,TOPKAT和TEST软件都显示出良好的预测能力,预测值与文献值的线性决定系数R~2为0.854~0.986(健康毒性)(P<0.01)、0.821~0.981(生态毒性)(P<0.01)。[结论]TOPKAT、TEST软件能高效、正确地预测大部分化学物的健康和生态毒性.可为新化学物的健康和生态毒性评价提供依据,为药物的早期开发提供决策支持。
[Objective] To summarize the predictability and applicability of TOPKAT and TEST softwares for the health and ecotoxicity of toxic chemicals, and to explore efficient, simple and rapid practical methods for toxicological detection. [Methods] The toxicology of chemicals was predicted by using SMILES combined with two toxicity prediction softwares. The prediction results were compared with the toxicology data released by European Chemicals Agency. The two softwares were summarized The advantages and disadvantages. [Results] TOPKAT and TEST software were able to predict the most common chemical toxicity. TEST software had higher predicted chemical coverage than TOPKAT software, and could predict polysubstituted benzene rings, heterocycles and macromolecular chemicals. TOPKAT could not Predict the toxicity of pyridines, some phenols and aniline chemicals. The accuracy of the two toxicity prediction software in predicting whether a chemical is toxic or not is about 85%. The accuracy of TOPKAT software in predicting the toxicity of pesticide chemicals was higher than that of TEST software. The linear regression coefficients of TOPKAT predictive value and literature value were 0.852 (healthy) (P <0.01) and 0.826 (ecotoxicity) (P < 0.01). The results of TEST software for the prediction of toxicity of benzene-based chemicals are slightly better than that of TOPKAT software. The linear determination coefficient (R ~ 2) of TEST predictive value and literature value are 0.932 (P <0.01), 0.787 (Ecotoxicity) <0.01). For ester and alcohol chemicals, TOPKAT and TEST software all showed good predictive ability. The linear determination coefficient (R ~ 2) between predicted value and literature value ranged from 0.854 to 0.986 (healthy) (P <0.01), and from 0.821 to 0.981 (Ecotoxicity) (P <0.01). [Conclusion] The TOPKAT and TEST software can predict the health and ecotoxicity of most chemicals efficiently and correctly, and provide a basis for the evaluation of the health and ecotoxicity of new chemicals, and provide decision-making support for the early development of drugs.