摘要
以自组建的食源性血管紧张素转化酶(angiotensin I-converting enzyme,ACE)抑制二肽为研究样本,采用氨基酸描述子VHSE(principal component score vector of hydrophilicity,steric,and electronic properties)对ACE抑制二肽进行表征后,比较偏最小二乘(partial least square regression,PLS)、支持向量机(support vector machine,SVM)及主成分分析(principal component analysis,PCA)-SVM相结合的3种建模方法对ACE抑制二肽的QSAR(quantitative structure-activity relationship)建模。结果显示,对于食源性ACE抑制二肽,3个模型的拟合能力无明显差异,SVM模型的预测能力略强;对其进行权重投影分析发现,C末端氨基酸较N末端氨基酸对其活性的影响更为明显。
A new ACE inhibitory dipeptides database was self-established. After the structures of dipeptides were characterized by using amino acid descriptors VHSE (principal component score vector of hydrophilicity, steric, and electronic properties), three kinds of modeling methods, namely partial least square regression (PLS), support vector machine (SVM), and principal component analysis(PCA) combined with SVM were used to establish the models of the QSAR of ACE inhibitory dipeptides, respectively. The results showed that there is no significant difference between the fitting abilities of the three models; the predictive abilities ofSVM model are stronger than other models. Moreover, the key structure factors relevant with dipeptide activities were studied. The results showed that the effect of the amino acid at C-terminal on ACE inhibitory activity of the dipeptides is more obvious than that at N-terminal.
出处
《浙江科技学院学报》
CAS
2015年第3期174-182,共9页
Journal of Zhejiang University of Science and Technology
基金
浙江省自然科学基金项目(LQ12C19004)
关键词
血管紧张素转化酶
二肽
支持向量机
偏最小二乘
定量构效关系
angiotensin I-converting enzyme(ACE)
dipeptide
support vector machine(SVM) partial least square regression(PLS)
quantitative structure-activity relationship(QSAR)