HashESU一种生物网络模体识别高效方法.pdfVIP

小型微型计算机系统 2015 年9 月第9 期 Joumal of Chinese Computer Systems Vol. 36 No.9 2015 HashESU: 一种生物网络模体识别高效方法 赵静，钟诚 (广西大学计算机与电子信息学院，南宁 53(阳的 E-mail:chzhong@ 摘 要:通过模体识别可以获得生物网络结构和功能，生物网络模体识别过程涉及到子图枚举和子图同构的问题，计算量非常 大.提出一种高效的网络模体识别算法 Has回SU，它使用经典算法 ESU 枚举子圈，采用 Hash 农结构保存生成的子圈，利用每个 新增结点与已经确定的结点之间的关系，生成子图标识关键字 SIK 并映射到 Hash 表中，每个 SIK 只在第一次生成时才需要调 用同构计算，以大大减少调用 NAUTY 算法进行同构检测的次数、更快地进行查找和插入子图操作，进而加快模体识别的速度. 实验结果表明，在模体识别结果质量相同的前提下，Has回SU 算法的运行效率明显优于著名的 ESU 算法和使用四分树结构存 储子围的QuateXelero 算法. 关键词:生物网络;模体识别;子图枚举;子图同构;Hashing 中图分类号:T凹01 文献标识码:A 文章编号:1侧-1220(2015) 剧创2-05 HashESU :Efficient Algorithm for Identifying Motifs in Biologiω1 Networks ZHAO Ji吨，ZHONG Cheng (Scb∞1 of Computer and Electronics and Information ,Guangxi University ,N缸ming 53创剧，China) Abstract :Network motifs identification is helpful for understanding the structure and function of biological netwo汰s.ldenti勿ing the motifs using subgraph enumeration and subgraph isomorphism is intensive-computation and time-consuming. An efficient networks motifs identification algorithm HashESU is proposed. HashESU uses classical algorithm ESU to enumerate 由e sub-graphs ,constructs hash table structure to sωife the sub-gr西phs and uses 由e relationship between the new node and old one in the subgraph to generate a subgraph identification key SIK which be map严d to the hash table. Generating SIK at the first time only calls isomorphism algorithm. η白白nes to call NAUTY isomorphism algori也m can be greatly reduced and the subgraphs 缸e fast searched and insertedωaccelerate the motif identification. Experimental results show that on 由e prernise of the motif identification results with the s缸ne quality ,the exe- cution efficiency of HashESU algorithm is superior to ESU algorithm and QuateXelero algorithm using quatemary tree