Based on the amino acid sequence conservation of Protein Fold Pattem. Protein fold pattem are predicted using amino acid composition and Polarity of amino acid, Friend or stranger river character of amino acid and Electric properties of amino acid . Moreover, using "one - versus - others" strategy and constructing position weight matrix and sequence pattern segment, Protein Fold Pattern is recognized by five similarity scoring functions. The best prediction accuracy can reach 83. 46% . The results show that the scoring matrix is a effective method for mtulti - class protein fold prediction.%依据蛋白质折叠子中氨基酸保守性,以氨基酸、氨基酸的极性、氨基酸的电性以及氨基酸的亲-疏水性为参数,从蛋白质的氨基酸序列出发,采用"一对多"的分类策略,通过构建打分矩阵和选取氨基酸序列模式片断,利用5种相似性打分函数对27类折叠子进行识别,最好的预测精度达到83.46%.结果表明,打分矩阵是预测多类蛋白质折叠子有效的方法.
展开▼