基于偏最小二乘法的乌头碱类化合物毒性的定量构效关系研究  

Study on the quantitative structure-toxicity relationships of aconitine compounds basing on partial least squares method

作　　者：李佐静[1] 王琳[2] 闫心丽[1] 高志祥[2] 孟繁浩[2] 

机构地区：[1]沈阳药科大学基础学院,辽宁沈阳110016 [2]中国医科大学药学院,辽宁沈阳110001

出　　处：《计算机与应用化学》2011年第6期765-768,共4页Computers and Applied Chemistry

基　　金：国家自然科学基金资助项目(30571591)

摘　　要：乌头碱类化合物属于二萜类生物碱,存在于乌头属欧乌头、川乌、北草乌和华乌头等多种毛茛科植物中,该类化合物既是其活性成分,也是其毒性成分。利用化合物定量结构-毒性效应关系（QSTR）方法研究了14个乌头碱类化合物的各种量化参数对其毒性的影响,并建立了毒性预测模型。由于该类化合物是从中药中提取分离的生物碱,样本数量相对较少,本文采用了偏最小二乘回归方法（PLS）进行降维,进行4个成分的提取及建模。毒性预测模型为：Log（toxi）=0.1593^＊Mass＋0.2908^＊LogP＋1.5475^＊SAA-0.5222^＊SAG-0.6104^＊Volume＋0.3112^＊Ref＋0.1784^＊Polar＋0.1785^＊BE＋0.1634^＊HF-0.1387^＊Dipole＋0.1412,结果表明该模型具有较好的毒性预测能力。Aconitine compounds are found in Aconitum Ouwu head,Chuan Wu,Aconitum,and the North China and other Ranunculaceae Aconitum plants,which are not only the active ingredient but also the toxic component.The quantum chemistry parameters of 14 kinds of aconitine compounds were studied using the quantitative structure toxicity relationships（QSTR）method to affect its toxicity and to build the forecast model.In this paper, Partial least squares method was taken into account to reduce the dimension and build the model using 4 principle components because of the limitation of sample space.It is tested that the model has a good forecasting ability.The model is：Log（toxi）=0.1593^＊Mass＋0.2908^＊LogP＋1.5475^＊SAA-0.5222^＊SAG-0.6104^＊Volume＋0.3112^＊Ref＋0.1784^＊Polar＋0.1785^＊BE＋0.1634^＊HF-0.1387^＊Dipole＋0.1412.

关 键 词：乌头碱类化合物 构效关系 偏最小二乘法 毒性预测 

分 类 号：O641[理学—物理化学]