论文部分内容阅读
环糊精在药剂学领域中是一类重要的包结化合物,其中络合物稳定常数(log K)是一个关键评价参数.本研究基于扩展距离矩阵提出了一组范数指数,利用多种计算方法构建了系列定量构效关系模型,并对233种化合物与β-环糊精的log K进行了计算预测.计算结果表明基于扩展距离矩阵范数建立的系列定量构效关系模型均能较好预测log K;其中利用最小二乘-支撑向量机方法建立的模型预测效果最好,其预测结果的相关性系数R和留一、留十交叉验证相关性系数(QLOO,QLTO)分别为0.9587、0.8775和0.8732.与文献方法对比结果表明,本工作的预测结果在准确性和稳定性上有着显著的改善,且能分辨同分异构体.本课题组前期研究成果和本项工作表明基于范数指数构建的定量构效关系不仅适用于计算化合物的基础物理化学性质,还能应用到化学反应过程相关常数的准确预测.
Cyclodextrin is a kind of important inclusion compound in the field of pharmacy, in which the log K of the complex is a key evaluation parameter.In this paper, a set of norm index is proposed based on extended distance matrix, A series of quantitative structure-activity-relationship models were constructed and the log K of 233 compounds and β-cyclodextrin were calculated and calculated.The results show that the series of quantitative structure-activity relationship models based on extended distance matrix norm can all be compared The prediction of log K is the best. The model predicted by least-squares support-vector machine method is the best one, and the correlation coefficient R and the remaining one of the prediction results, QLOO and QLTO are 0.9587 , 0.8775 and 0.8732 respectively.Compared with the literature methods, the results show that the prediction results of this work have significant improvements in accuracy and stability and can distinguish isomers.The preliminary research results and this work show that this work is based on Quantitative structure-activity relationship constructed by norm index is not only applicable to calculate the basic physical and chemical properties of compounds, but also can be applied to the accurate prediction of the relevant constants of the chemical reaction process.