蛋白质折叠类型识别方法研究

蛋白质折叠类型识别是一种分析蛋白质结构的重要方法.以序列相似性低于25%的822个全B类蛋白为研究对象,提取核心结构二级结构片段及片段问氢键作用信息为折叠类型特征参数,构建全B类蛋白74种折叠类型模板数据库.定义查询蛋白与折叠类型模板间二级结构匹配函数SS、氢键作用势函数BP及打分函数P,P值最小的模板所对应的折叠类型为查询蛋白的折叠类型.从SCOP1.69中随机抽取三组、每组50个全β类蛋白结构域进行预测,分辨精度分别为56%、56%和42%;对Ding等提供的检验集进行预测,总分辨精度为61.5%.结果和比对表明,此方法是一种有效的折叠类型识别方法.

Study on protein fold type recognition

A fold type recognition method is established based on the One of the important approach to Protein structure analysis is protein fold type recognition. Characteristics including secondary structures and hydrogen bonds between them in the structure core are drawn from 822 all-βdomains with sequence identity below 25%, and thus constitute an non-redundant library of all-βprotein that includes 74 fold types. Secondary structure alignment function SS and long-range contact potential BP are defined, score function P is developed to fold type recognition. The predicted fold type is the entry with minimal P value. For randomly picked three test sets with 50 entries in each, the fold prediction accuracy are 56%, 56% and 42% respectively; As to test set provided by Ding, the total fold prediction accuracy is 61.5%. The results and comparison demonstrate the effectiveness of FRESH.