基于连接强度的动态蛋白质网络构建算法研究  

Research on Dynamic Protein Network Construction Algorithm Based on Connection Strength

作　　者：李鹏 闵慧[4] 罗爱静 许家祺[1] 颜湘茹 伊娜 刘杰 LI Peng;MIN Hui;LUO Aijing;XU Jiaqi;YAN Xiangru;YI Na;LIU Jie(School of Informatics,Hunan University of Chinese Medicine,Changsha 410208;The Third Xiangya Hospital of Central South University,Changsha 410006;Key Laboratory of Medical Information Research(CSU),College of Hunan Province,Changsha 410006;Software Institute,Hunan College of Information,Changsha 410200)

机构地区：[1]湖南中医药大学信息科学与工程学院,长沙410208 [2]中南大学湘雅三医院,长沙410006 [3]医学信息研究湖南省普通高等学校重点实验室(中南大学),长沙410006 [4]湖南信息职业技术学院软件学院,长沙410200

出　　处：《计算机与数字工程》2020年第11期2539-2543,共5页Computer & Digital Engineering

基　　金：国家社会科学基金重点项目(编号:17AZD037);国家重点研发计划(编号:2017YFC1703306);湖南省自然科学基金青年项目(编号:2019JJ50453);湖南省自然科学基金面上项目(编号:2018JJ2301);湖南省科技厅重点项目(编号:2018JJ2301);湖南中医药大学开放基金项目(编号:2018JK02);湖南省教育厅一般项目(编号:19C1318)资助。

摘　　要：为了解决现有的蛋白质网络构建方法在准确性、鲁棒性等方面的不足,提出了一种基于连接强度的动态蛋白质网络构建算法。首先,基于基因表达均值计算来对所有的蛋白质划分时间片,然后对于拥有相同时间片的蛋白质,采用连接强度来判断它们之间是否存在相互作用关系,进而构建出相应的蛋白质子网。最后,基于进化图来将多个时间片内的子网构建成动态蛋白质网络模型。以DIP数据集和CYC2008数据集为实验对象,利用多种典型的复合物识别算法在文中构建的蛋白质网络上进行测试,结果表明,论文构建的蛋白质网络明显地提高了复合物识别的精度,且鲁棒性较高。In order to solve the shortcomings of existing protein network construction methods in accuracy and robustness,a dynamic protein network construction algorithm based on connection strength is proposed.Firstly,all proteins are divided into time slices based on the calculation of mean gene expression.Then,for proteins with the same time slice,the interaction between them is judged by the connection strength,and then the corresponding protein sub-network is constructed.Finally,a dynamic protein net⁃work model is constructed based on evolutionary graph.In this paper,DIP data set and CYC2008 data set are used as experimental objects,and a variety of typical complex recognition algorithms are used to test on the protein network constructed in this paper.The results show that the protein network constructed in this paper significantly improves the accuracy of complex recognition and has high robustness.

关 键 词：动态蛋白质网络 连接强度 时间片 进化图 精度 鲁棒性 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]