基于奇异值分解的RNA二级结构相似度计算方法  

Computing method of similarity between RNA secondary structures based on singular value decomposition

作　　者：刘琦[1] 张引[2] 叶修梓[2] 俞荣栋[1] 

机构地区：[1]浙江大学沃森基因组科学研究院 [2]浙江大学计算机科学与技术学院,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2007年第8期1249-1254,共6页Journal of Zhejiang University：Engineering Science

摘　　要：针对传统的核糖核酸(RNA)分子结构相似度计算方法存在复杂度高、未能充分考虑二级结构全局信息的缺点,基于一种7维序列编码方法,提出了一种利用矩阵奇异值分解提取RNA二级结构数据分布特征的算法,利用提取出的奇异值向量,基于向量距离进行RNA二级结构相似度比对.通过对Yeast tRNA-phe、PSTV、poliovirusmutant 3种RNA分子二级结构构象的实验分析,表明了该方法的可行性和高效性.并将这种方法推广至任意两个RNA分子之间二级结构的相似度衡量,同样取得了不错的效果.Traditional methods of comparing the similarity between Ribonucleic Acid（RNA） structures have high computing complexity and consider insufficiently about the overall structure information.A method of extracting the distribution feature of RNA secondary structures based on the 7-D encoding and the singular value decomposition was presented.The similarity was compared by computing the distance based on the vectors of singular values.Experiment on the ensemble structures of three kinds of RNA： Yeast tRNA-phe,PSTV,poliovirus mutant,demonstrated the feasibility and effectiveness of the approach.Finally the method was generalized to the structure similarity comparison between random RNA molecules,and acceptable results were obtained.

关 键 词：RNA二级结构 相似度 奇异值分解 特征提取 

分 类 号：Q71[生物学—分子生物学] TP181[自动化与计算机技术—控制理论与控制工程]