论文部分内容阅读
应用分子力学方法MM~+和半经验量子化学AM~1法几何全优化得到20个己烷雌酚衍生物的优势构象,再利用量子化学算法和分子图形学技术获得电子结构参数和几何结构参数,采用多元逐步回归分析和人工神经网络误差反传算法(BP)研究衍生物与雌激素受体相对亲和力的QSAR。结果表明:分子最低空轨道能、7号碳原子和10号碳原子间键长有利于提高亲和力;而引入大体积取代基对化合物亲和力的提高不利,建立了20个己烷雌酚衍生物的构效关系式,所建方程的相关系数及去一法交互检验复相关系数分别为0.963和0.880。
The dominant conformations of 20 hexestrol derivatives were obtained by using the molecular mechanics method MM ~ + and semi-empirical quantum chemistry AM-1 method. The electronic structure parameters and geometrical structure parameters were obtained by using quantum chemical algorithm and molecular graph technique The QSARs were used to study the relative affinity between the derivatives and the estrogen receptor by multiple stepwise regression analysis and artificial neural network error back propagation algorithm (BP). The results showed that the lowest orbital energies of the molecules, the bond length between the carbon atom 7 and the carbon atom 10 could improve the affinity, while the introduction of the bulky substituents had an adverse effect on the increase of the affinity of the compound. The structure-activity relationship, the correlation coefficient of the built equation and the reciprocal test-to-law correlation coefficient are 0.963 and 0.880 respectively.