论文部分内容阅读
将虚拟受体的概念与比较分子场法有机结合起来,得到了一种新颖的3D-QSAR研究方法:比较分子／虚拟受体相互作用分析法(CoMPIA)．该方法通过定义受体原子探针并利用遗传算法优化探针在药物分子周围的最佳分布情况来找寻模型最优解和虚拟受体模式,从而有效提高了建模的质量和结果的可解释性．使用CoMPIA对31个经典甾体化合物进行了系统的QSAR研究,所得模型拟合复相关系数R2、交叉检验Q2以及对测试集预测结果的均方根误差RMSEF,分别为0．940,0．868和0．502,该结果优于多数文献报道．
By combining the concept of virtual acceptor with the comparative molecular field method, a novel 3D-QSAR method was developed: the comparative molecular / virtual receptor interaction analysis (CoMPIA). The method can find the optimal model and the virtual receptor model by defining the receptor atom probe and using the genetic algorithm to optimize the best distribution of the probe around the drug molecule, so as to effectively improve the quality of modeling and the interpretability of the result Sex. A systematic QSAR study of 31 classical steroids was performed using CoMPIA. The resulting model fitted the complex correlation coefficient R2, the cross-test Q2, and the root-mean-square-root-mean-square error RMSEF for the test set, respectively, at 0.940 and 0.868, respectively And 0.502, which is better than most reports.