改进的离散增量算法预测27类折叠子的结构类型  

An Improved Method for Predicting Structure Class of 27-Class Protein Folds Using Increment of Diversity

作　　者：张怀光[1] 胡秀珍[1] 

机构地区：[1]内蒙古工业大学理学院,呼和浩特010051

出　　处：《内蒙古大学学报（自然科学版）》2009年第3期285-290,共6页Journal of Inner Mongolia University：Natural Science Edition

基　　金：内蒙古自治区高等学校科学研究项目(NJZY08059)

摘　　要：蛋白质二级结构预测是三级结构预测的一个非常重要的中间步骤,而折叠子识别和结构类型的准确预测则可以提高二级结构和三级结构预测的准确度.本文从蛋白质的一级序列出发,提出了一种改进的预测算法:以二肽组分、预测的二级结构信息、伪氨基酸组分和位置权重矩阵打分值等特征分别作为参数,输入离散增量算法的单分类器中,通过加权融合单分类器的计算结果,对27类折叠子的结构类型进行了预测,取得了较好的预测结果.The protein secondary structure recognition is an important step to search protein tertiary structure, and the recognition of protein folds and structure class may enhance the prediction accuracy of secondary structure and tertiary structure. According to the primary structure of proteins,an improved method for predicting structure class of 27-class protein folds is proposed. Based on the increment of diversity algorithm,a set of classifieries is built by different characteristic parameters, such as dipeptide composition, predicted secondary structure, pseudo amino acid composition and position weight matrix score. Then the outcomes of various classifieries are fused with weighted factors,and a better result is obtained.

关 键 词：离散增量 伪氨基酸组分 位置权重矩阵 蛋白质折叠子 蛋白质结构类型 

分 类 号：Q61[生物学—生物物理学]