摘要
为了研究羟酸类和氨基酸类手性化合物的构效关系,本实验引进了手性乘积算法:P=(1-b/a)(1-c/b)(1-a/c)(1-a/d),其中a、b、c和d分别为与分子中手性中心碳原子4个不同连接片断手性eAm指数和分子连接性指数mx的计算值,并考察了所得参数与羟酸类和氨基酸类化合物手性薄层色谱保留指数的相关性。为了评定所得预测模型的稳定性,文中运用了交叉验证法。结果表明,由人工神经网络方法(BP法)所得数学模型比较稳定,且明显优于多元回归分析结果。
For the studies on quantitative structure-activity/property relationships (QSAR/QSPR) of hydroxyl acids and amino acids, the algorithm of chirality product P was introduced : P = ( 1 - b/a) ( 1 - c/b) ( 1 - a/c) (1 -a/d) , where a, b, c, and d are the values of chirallity indices eAm and m^x. To take the values derived based on above indices as the parameters, the relationships between the structures of 16 hydroxyl acids and amino acids and their chiral high-pressure thin-layer chromatographic data have been observed. For verifying the model obtained by using artificial neural networks, the method of cross-validation has been used. The results show that the prediction model is stable, and much better than that obtained by using multiple regression analysis.
出处
《分析化学》
SCIE
EI
CAS
CSCD
北大核心
2009年第4期543-547,共5页
Chinese Journal of Analytical Chemistry
关键词
手性化合物
构效关系
人工神经网络
多元回归分析
交叉验证
Chiral compounds, quantitative structure-activity/property relationship, artificial neural networks, multiple regression analysis, cross-validationj