基于基团属性的手性Am指数及应用

Chiral Am index based on properties of ligands and its application

代替原子序数,采用有意义的属性,例如基团尺寸,来比较手性中心键连4个基团的大小,并通过各基团属性的排序判断手性中心的构型为类R/S。以所得到的新构型为基础,扩展了基于分子结构和距离矩阵的Am指数,提出了新的手性Am指数,以其与物化手性描述符相结合来表征一个分子的手性,并应用于脂化酶作为催化剂条件下,手性伯醇产物的立体选择预测以手性指数和Random Forests建立模型,通过交叉验证,能正确预测整个数据集的90%的对映异构体。

The four ligands of chiral center were ranked according to meaningful property, such as size of the ligands, instead of atomic number, and the configuration of chiral center was assigned to an S/R-like relative to that property. Am indices derived from the molecular structures and distance matrixes were extended on the basis of the new configuration, and the new chiral Am indices, which combined which physicochemical atomic stereodescriptors, were implemented to represent the chirality of a molecule. The chiral Am indices were used to investigate the enantiopreference of Pseudomonas cepaeia lipase （PCL） toward chiral primary alcohols. The relationship between the chiral descriptors and the enantioselectjvity was learned by Random Forest. The learned models could make correct prediction of the preferred enantiomers to 90% of the whole data set by cross validation.