基于拓扑和生物特征的权重网络中络合物抽取  

Complex Extraction from the Weighted Network Based on Topological and Biological Characteristics

作　　者：于凤英[1] 杨志豪[1] 林鸿飞[1] 

机构地区：[1]大连理工大学计算机科学与技术学院,辽宁大连116024

出　　处：《江西师范大学学报（自然科学版）》2013年第3期273-278,共6页Journal of Jiangxi Normal University(Natural Science Edition)

基　　金：国家自然科学基金(61272373;61070098;60973068);国家"863"计划(2006AA01Z151);中央高校基本科研业务费专项基金(DUT10JS09)资助项目

摘　　要：蛋白质关系网络的权重设置对蛋白质络合物的抽取有着较大的影响.综合考虑蛋白质关系网络的拓扑结构特征和生物信息特征,提出一种新的权重设置策略,并向原有蛋白质关系网络中添加高可信度关系.基于修正后的蛋白质关系网络,用监督学习方法抽取蛋白质络合物,在DIP数据集上,F值达到0.570 5.The weight setting of protein interaction network has a great effect on protein complex identification.A better strategy which considers both topological characteristics and biological characteristics of protein interaction network has been provided.Furthermore,some credible interactions have been added into the original network.Then,the updated protein interaction network has been used for complex identification based on a supervised method.F-score reached 0.570 5 in the DIP dataset.

关 键 词：权重网络 权重设置 蛋白质关系 蛋白质络合物 监督学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]