基于电负性构象独立手性指数及应用

Conformation-independent chirality code based on electronegativity and its applications

为了采用计算方法研究手性化合物,为了通过结构-活性相关性研究预测与手性有关的性质,本文提出了采用σ和π的残余电负性之和作为原子属性的构象独立手性指数。该手性指数衍生于原子径向分布函数,包含分子几何和原子属性的信息,能够区分对映体。将构象独立手性指数应用于1个包含48对手性氨醇对映体的数据集,该数据集为苯甲醛与二乙基锌发生加成反应的催化剂,每个催化剂均产生特定绝对构型的反应主产物。采用相向传输神经网络建立了手性氨醇催化剂的构象独立手性指数与反应主产物绝对构型的相关性模型,得到了满意的预测结果。对于独立的测试集,90.0%的催化剂被正确地预测;对于训练集,89.5%的催化剂被正确地识别。

In order to investigate chiral compounds by computational methods and to obtain predictions by studies on structure-activity relationship of chiral compounds for properties that are influenced by chirality,conformation-independent chirality code with sum of pi and sigma residual electronegativity as atomic property was implemented in this article.The chirality codes are derived from atomic radial distribution function,composed of information about molecular geometry and atomic property,and able to distinguish between enantiomers.The chirality codes were applied to a data set of 48 enantiomeric pairs of chiral amino alcohols that enantioselectively catalyze the addition of diethylzinc to benzaldehyde.Each amino alcohol yields preferential absolute configuration of primary product.The relationship between chirality code of amino alcohol and the absolute configuration of primary product was built by counter-propagation neural network,and the results were satisfactory.For independent test sets,90.0%of catalysts were correctly predicted,and 89.5%were correctly recognized for training set.