论文部分内容阅读
应用MM3分子力学方法对R6=H的13个紫杉醇类似物进行几何优化,并采用MNDO法计算了化合物的电子结构,应用回归分析、神经网络等方法寻找其量化指数与抗癌活性的关系.结果表明,R2取代基、2-OBz中苯甲酰基(Bz)的碳氧原子的净电荷之和∑QBz、核-核排斥能Ep对活性影响较大.并得到显著性较好的QSAR:ID50(A)/ID50(T)(act)=71.20535+155.35016∑QBz,表明随着∑QBz减小,ID50(A)/ID50(T)值将减小,而化合物活性将增大.神经网络(BP网络)算法的结果表明,神经网络计算值R=O.992,神经网络的结果可精确地预测化合物的抗癌活性.
The 13 paclitaxel analogues of R6=H were geometrically optimized using MM3 molecular mechanics method, and the electronic structure of the compound was calculated using the MNDO method. The relationship between the quantitative index and the anticancer activity was investigated using regression analysis, neural network and other methods. The results show that the R2 substituent, the sum of the net charge of the carbon and oxygen atoms of benzoyl (Bz) in 2-OBz, ∑QBz, and the nuclear-nuclear exclusion energy Ep have a great influence on the activity, and the QSAR with better significance is obtained: ID50 (A)/ID50(T)(act) = 71.20535 + 155.35016 ∑ QBz, indicating that as ∑QBz decreases, the ID50(A)/ID50(T) value will decrease and the compound activity will increase. Neural networks ( The results of the BP network) algorithm show that the neural network calculated value is R=O.992. The results of the neural network can accurately predict the anticancer activity of the compound.