论文部分内容阅读
目的为了寻找新一代紫杉醇类抗癌药物,研究紫杉醇类似物分子结构和抗癌活性(PIDact)之间的定量构效关系。方法基于电性状态拓扑指数和电性距离矢量,应用最佳子集回归的方法建立了PIDact值和紫杉醇类似物分子结构的定量结构–活性相关(QSAR)模型,并对模型进行了交互检验和外部检验,用43个紫杉醇类似物训练集样本构建的QSAR模型预测了外部10个检验集的PIDact值。结果所建立模型具有较高的估计相关系数及LOO(leave-one-out)检验相关系数。结论模型具有良好估计能力与稳定性,训练集模型具有良好预测能力。
In order to find a new generation of paclitaxel anti-cancer drugs, the quantitative structure-activity relationship between the molecular structure of paclitaxel analogs and the anti-cancer activity (PIDact) was studied. Methods Quantitative structure-activity correlation (QSAR) models of PIDact value and paclitaxel analogue molecular structure were established by the best subset regression method based on the electrical state topological index and electrical distance vector. The model was interactively tested and Externally, the QSAR model constructed with 43 paclitaxel analog training samples predicts PIDact values for the 10 external test sets. Results The model established had higher estimated correlation coefficient and LOO (leave-one-out) test correlation coefficient. Conclusion The model has a good estimation ability and stability, and the training set model has good predictive ability.