中文摘要：

蛋白质折叠类型分类是蛋白质分类研究的重要内容。以SCOP数据库中的PHdomain—likebarrel折叠类型为研究对象，选择序列相似度小于25％的61个样本为检验集，通过结构特征分析，确定了该折叠类型的模板及其对应的特征参数，利用模板与待测蛋白的空间结构比对信息，提出了一个新的折叠类型打分函数Fscore，建立了基于Fscore的蛋白质折叠类型分类方法并用于该折叠类型的分类。用此方法对Astrall．75中序列相似度小于95％的16711个样本进行检验，分类结果的特异性为99．97％。结果表明：特征参数抓住了折叠类型的本质，打分函数Fscore及基于Fscore建立的分类方法可用于PH domain—like barrel蛋白质折叠类型自动分类。

英文摘要：

Folding is an important issue for protein classification. Based on the classification of SCOP1.75, 61 proteins （ 〈25% sequence identity） of PH domain-like barrel are selected, we get the template of folding types and the corresponding characteristic parameters by analysing the structure. Using the contrast information between the template and the spatial structure of the test proteins, this paper presents a new scoring function Fscore. Establishes a protein folding type classification method based on Fscore and applies it to this folding type classification. In Astrall. 75 database, we select 16712 proteins （ 〈95% sequence identity） as the testing sample, the sensitivity is 99.97%. The result indicates that characteristic parameters describe the nature of the folding type, the scoring function Fscore and the classification method based on Fscore can be used for automatic classification of PH domain -like barrel protein folding type.