中文摘要：

蛋白质间的相互作用在信号转导和其他许多生物进程有着至关重要的作用。从20种天然氨基酸的554个物化性质中,单变量建模方法挑选出疏水、静电、立体、氢键4个描述子表征能与人类双载蛋白SH3结构域结合的多肽,预测SH3结构域-肽亲和力。所得描述子并未对变量进行主成分分析（PCA）压缩,且只对氨基酸侧链性质进行结构表征,因此应用这4个描述子并结合多元线性回归（MLR）建模方法对SH3结构域,肽体系进行定量构效关系（QSAR）研究分析域,肽亲和力。测试集的多肽用于模型的验证,内部验证复相关系数（R~2）和外部留一法交互验证相关系数（Q~2）分别为：0.682,0.650,预测均方根误差（RMSE）为0.528,从模型统计数据可知,QSAR模型预测模型稳定性高,预测能力强。说明这4个描述子物具有更为确切的物化意义,表征过程更加简洁有效且研究效率高的多重优点,并可以拓展预测不同的域,肽体系。

英文摘要：

Protein-protein interaction play critical roles in signal transduction and may other key biological process.In this article,we discuss the application of the quantitative structure-activity relationship（QSAR） for the modeling and prediction of binding affinities between the human amphiphysin SH3 domains and its peptides ligands.This four descriptors,which have been the key factors to influence of amino acid residues and don＇t used to principal component analysis（PCA）,selected form 554 physiochemical variables of 20 natural amino acids separately according to different kinds of properties descried,namely,hydrophobicity,electronic,steric property,hydrogen bond.They were applied to structure characterization and QSAR analysis on SH3 domains and its peptide ligands Parameters being responsible for the binding affinities were selected by single-variable modeling,and a QSAR model by multivariate linear regression analysis（MLR） methods to predict the peptides-SH3 domains interaction.The leave one out cross validation values（Q~2）,the multiple correlation coefficients（R~2） and the root mean squares error（RMSE） is 0.650, 0.682,0.528,respectively.Test sets of peptides were used to validate the quantitative model,and it was shown that all these QSAR models had good predictability for outside samples.The result showed that,in the four descriptors can well represent the structural characterization of SH3 domains-decapeptides QSAR analysis,which has multiple advantages,such as define physical and chemical meaning,low,computational complexity, easy to get,and good structural characterization ability.It provides a framework to derive integrated prediction models for different domains-peptides systems.