摘要
采用比较分子场 (CoMFA)分析法和偏最小二乘回归 (PLS)法对十多种HMG -CoA还原酶阻滞剂的构效关系进行了多维定量建模研究。在描述变量数远远超过样本数的情况下 ,经典的Hansch方法等均不适应 ,而笔者的方法可给出良好结果。
Multidimensional multivariate quantitative structure-activity relationships (MD/MV QSAR) are moddeled for thirteen HMG CoA reductase inhibitors using partial least squares regression (PLS) and comparative molecular field analysis (CoMFA) modified by introducing molecular lipophilicity potential (MLP) as a new hydrophobic field. Under condition that the number (m) of variables is much greater than the number (n) of samples, m>>n , many techniques such as the classical Hansch multiple linear regression (MLR) method, are not suitable; whereas the developed methods based on the combination of the modified CoMFA and latent PLS modelling suitable and applicatable, which privides quite good results for calibration and prediction.
出处
《重庆大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2001年第4期154-158,共5页
Journal of Chongqing University
基金
国家新药基金 ( 96-90 1-5 -4 )
教育部春晖计划启动基金 ( 1999-3 8)
霍英东青年教师基金 ( 1998-0 3 -2 8)
日本政府文部省与
关键词
HMG-COA
多维多元定量构效关系
比较分子场分析
分子建模
药物
还原酶阻滞剂
multidimensional multivariate quantitative structure-Activity relationships
HMG CoA reductase inhibitors
multidimensional multivariate molecular modelling lead optimization partial Least squares(PLS) modified comparative molecular fi
