基于压缩氨基酸和支持向量机进行膜蛋白类型识别.pdfVIP

Vol.l 1 No .4 第 11 卷第4 期 生物信息学 2013 年 12 月 Chinese Joumal of Bioinformatics Dec. ,20 l3 doi: 10. 3969/j.issn.1672-5565.2013.04.05 基于压缩氨基酸和支持向量机进行膜蛋白类型识别 管翠萍，石晶，徐惠娟* (宁夏大学生命科学学院，宁夏银川 750021 ) 摘 要:膜蛋白是一类结构独特的蛋白质，是细胞执行各种功能的物质基础。根据其在细胞膜上的不同存在方式，主要分为 六种类型。本文利用压缩的氨基酸对原始膜蛋白序列进行信息压缩，再对压缩序列进行氨基酸组成和顺序特征的提取，最后 采用支持向量机构建分类模型。通过五叠交叉验证的结果表明，该方法对于六种膜蛋白的分类预测，准确度最高可达 981毛以 上，平均预测准确度在 85% 以上，可有效实现膜蛋白六种类型的划分，为进一步分析膜蛋白的结构和功能奠定基础。 关键词:膜蛋白;压缩氨基酸;支持向量机 申图分类号:R978. 1+6 文献标识码:A 文章编号:1672-5565(2013)-04-271-04 Prediction of membrane protein types based on compressed amino acids and support vector machine GUAN Cui-ping , SHI Jing , XU Hui才uan* (L庐 scierue school , NingXia University , NingXia YinChuan 750021 , China) Abstract: Memhrane proteins which hold a particular structure 缸它 the exact substance in cells to implement various functions. Six types of memhrane proteins were classified based on their different performances on cell memhrane. In this study , compressed amino acids were used to compress the original memhrance proteins se年lences ， and features in the form of single amino acid and dipeptide compositions were extracted from the compressed sequences. Finally , classifiers were developed using suppo时 vector machine (SVM). 咀le results demonstrated that this method could well predict the types of memhrane proteins as the accuracy rate of prediction is above 98% and 859毛 on average bωed on 5-fold cross- validation. This work will establish the basis for the further research of memhrane protein s structure and function. Keywords: Membrane Protein; Compressed Amino