摘要
使用从分子一级拓扑结构出发、结合分子中非氢原子电性和键连属性以及原子间相对距离的分子电性距离矢量(MEDV-B),对58个血管紧张素转化酶(ACE)抑制剂二肽和48个苦味(BT)二肽进行定量结构活性相关(QSAR)研究,用多元线性回归建立矢量描述子与活性观测值间相关模型,并用留一法交互校验(LOO-CV)检验其预测力,取得较满意的结果(ACE:n=58,m=10,R=0.894,Rcv=0.818;BT:n=48,m=10,R=0.947,R=0.898);再用逐步回归对变量进行筛选与优化,建立新模型,稳定性与预测力得到进一步改善(AACE:n=58,m=5,R=0.859,Rcv=0.824;BT:n=48,m=5,R= 0.931,Rcv=0.908)。结果表明:该矢量描述子可用于二肽结构表征与生物功能预测,且计算简便。
The molecular electronegativity distance vector (MEDV-B), which based on the primary structure and combined with the electronegativity, bonding attribution of atom and the distance between atoms in the molecule, is used to the quantitative structure activity relationship (QSAR) of two dipeptide panels which consisted of 58 angiotensin converting enzyme (ACE) inhibitor dipeptides and 48 bitter tasting (BT) dipeptides. The multiple linear regression (MLR) is used to model the relationship between vector descriptors of dipepetide and its observed value. The estimation performance and prediction capability of the model are tested by the leave one out cross validation (LOO-CV). The satisfactory results are gained (ACE: n =58, m = 10, R =0. 894, Rct. =0. 818; n =48, m = 10, R=0. 947, RCV = 0. 898). The stepwise multiple regression (SMR) is use to solve the linear overlap of MLR model and optimize and filter the variables. At last, the optimized mode is constructed and its estimation performance and prediction capability are improved (n=58, m=5, R=0.859, Rci,=0.824; n=48, m=5, R=0.931, RCV =0.908). Based on the investigation, the vector descriptor is fit to characterize the structure of dipeptides and predict their biological activity and has the merit good correction and simple calculation.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2005年第9期759-762,共4页
Computers and Applied Chemistry
基金
国家教委霍英东基金国家"春晖计划"教育部启动基金重庆应用基础研究基金 2001-3-6重庆大学创新基金资助项目
关键词
定量结构活性相关
分子电性距离矢量
结构表征
生物功能预测
逐步回归
quantitative structure activity relationship (QSAR), molecular electronegativity distance vector (MEDV-B), molecular structure characterization, prediction of biological activity, stepwise multiple regression (SMR)