变形虫穿孔肽及其类似物的结构表征与抗菌活性定量预测  

Structural characterization and antibacterial activity prediction for amoebapores and their analogs

作　　者：熊清[1] 王远强[2] 李志良[3] 

机构地区：[1]西南大学计算机与信息科学学院,重庆400715 [2]重庆工学院生物工程学院,重庆400050 [3]重庆大学生物工程学院,重庆400044

出　　处：《计算机与应用化学》2006年第10期1007-1012,共6页Computers and Applied Chemistry

摘　　要：本文提出新的分子结构描述子一按键分类的分子电性距离矢量(B-MEDV)来表征变形虫穿孔肽及其类似物的分子结构,以对人类病原体Candida albicans的最低生长抑制浓度(MIC)的对数值为抗菌活性指标,借助多元线性回归(MLR)和逐步回归(SMR)建立定量结构活性相关(QSAR)模型。结果表明：所得模型相当稳定且有较强的预测能力；5种表征方案最优模型的相关系数为：0．94303、0．91167、0．90443、0．86581、0．94477；交互检验的相关系数为：0．73653、0．84285、0．81535、0．74451、0．83030。通过对比分析5种表征方案发现,多肽结构的特征主要表现在氨基酸的残基上,可以用残基的矢量描述子来表征整个多肽的结构,从而使表征过程更为简洁有效。A novel set of structural descriptors, called the bond-association classified molecular electronegativity-distance vector was developed to characterize the structures of amoebapores and their analogs, and quantitative relationship between structure and antibacterial activity was established by multiple linear regression and stepwise multiple regression. The results showed the models constructed were provided with favorable estimation stability and predictive ability. The correlation coefficients of the best models were 0. 94303, 0. 91167, 0. 90443, 0. 86581 and 0. 94477 respectively for five research schemes, the correlation coefficients of CV were 0. 73653, 0. 84285, 0. 81535, 0. 74451 and 0. 83030 respectively. Analysis of performance for five research schemes indicated that the residue is the main structural feature in this peptide system, and residue characterization can be used to descript the entire peptide. This will make characterization simpler and more efficient.

关 键 词：变形虫穿孔肽 按键分类的分子电性距离矢量 定量结构活性相关 多元线性回归 逐步回归 

分 类 号：Q6[生物学—生物物理学]