论文部分内容阅读
应用TOPKAT和Derek平台预测已建立的984种中草药重要成分的毒性,用TOPKAT平台预测中草药重要成分的大鼠经口LD50、大鼠经口慢性LOAEL、致癌性(啮齿类)、致突变性(Ames试验)和大鼠经口发育毒性;Derek预测中草药重要成分的致癌性、致突变性和染色体损伤。TOPKAT预测的毒性数据报告为连续数据和分类数据,分类数据用0.0~1.0的数值表示。LD50和慢性LOAEL根据GHS进行分类。TOPKAT预测结果:有866种(88%)中草药重要成分的大鼠经口LD50属GHS急性经口毒性3类、4类和5类;分别有214(21.7%)和253种(25.7%)中草药重要成分的慢性LOAEL属GHS慢性毒性1类和2类;有361种(36.7%)预测具有致癌性阳性,有206种(21%)预测具有Ames试验阳性,549种(55.8%)预测具有发育毒性。Derek使用9种术语由低到高定性描述预测结果的不确定性。Derek预测结果:有57.8%的中草药成分无结果,有131种(13.3%)中草药重要成分有致癌性结构预警,有82种(0.08%)有Ames试验结构预警,有245种(24.9%)有染色体损伤结构预警,对上述预测阳性结果描述不确定性的术语90%以上为肯定、很可能和合理的。TOPKAT和Derek是较好的商业QSAR平台,但对毒性终点的定义不完全相同,应用域也不一致,有待于进一步评价和优化。
The TOPKAT and Derek platforms were used to predict the toxicity of 984 established Chinese herbal medicines. The TOPKAT platform was used to predict the oral LD50 of rats, the chronic oral LOAEL of rats, carcinogenicity (rodents) and Ames Test) and oral developmental toxicity in rats; Derek predicts the carcinogenicity, mutagenicity and chromosomal damage of important components of Chinese herbal medicine. Toxicity data predicted by TOPKAT are reported as continuous data and disaggregated data, and disaggregated data are expressed as 0.0 to 1.0. LD50 and chronic LOAEL are classified according to GHS. TOPKAT Prediction Results: Oral LD50 of 866 species (88%) of major Chinese herbal ingredients were classified as GHS acute oral toxicity categories 3, 4 and 5; 214 (21.7%) and 253 (25.7%) Chinese herbal medicines Chronic LOAELs, which are important components, are chronic GCC class 1 and 2; 361 (36.7%) predict carcinogenic, 206 (21%) predict Ames positive and 549 (55.8%) predict developmental toxicity. Derek uses the nine terms from low to high to qualitatively describe the uncertainty of the predicted result. Derek predicts that there are 57.8% Chinese herbal ingredients with no result, 131 (13.3%) Chinese herbal ingredients have carcinogenicity early warnings, 82 (0.08%) have Ames early warnings and 245 (24.9%) have Chromosomal damage early warning structure, the positive predictive results described above to describe the uncertainty of more than 90% of the affirmative, it is possible and reasonable. TOPKAT and Derek are good commercial QSAR platforms, but the definitions of toxicity endpoints are not exactly the same, and the application domains are inconsistent, which needs further evaluation and optimization.