基于异构网络拓扑数据的人类必要基因预测  

Human essential gene prediction based on heterogeneous network topology data

作　　者：李建伟 岳宗河 黄焱 段向欢 LI Jianwei1＇2, YUE Zonghe1, HUANG Yan1, DUAN Xianghuan1(1. School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, China; 2. Hebei Province Key Laboratory of Big Data Calculation, Tianjin 300401, China)

机构地区：[1]河北工业大学人工智能与数据科学学院,天津300401 [2]河北省大数据计算重点实验室,天津300401

出　　处：《河北工业大学学报》2018年第3期36-41,共6页Journal of Hebei University of Technology

基　　金：国家自然科学基金(81672113)

摘　　要：对必要基因进行研究不仅能够了解生物生存和繁殖的最低要求,且有助于寻找人类疾病基因和新的药物靶点.实验法鉴定人类必要基因虽有效但价格昂贵且耗时费力,开发高效算法预测必要基因是对实验法必要而有效的补充.提出一种基于融合多个异构网络拓扑数据预测必要基因的算法,该算法选用重启动随机游走算法将多个异构网络整合成统一的基因网络特征,采用SMOTE过抽样算法平衡训练支持向量机过程中的正负样本.实验结果表明,整合异构网络拓扑数据方法比基于单一网络的模型能更有效地预测人类必要基因.The studies of the essential genes are helpful not only in understanding the mininmnl requirements for surviv- al and reproduetion, but also finding the new- human disease genes and drug targets. Though the experimental methods to identify the essential genes is effeetive, these methods are expensive and time-eonsunling. Therefore, the development of effieient predietion algorithm to prediet human essential genes is a neeessary and effeetive eomplement to experimental methods. This paper proposed an algorithm based on the fusion of multiple heterogeneous network topology data to prediet the essential genes. In our study, random walk with restart algorithm was used to integrate heterogeneous network topo- logieal data into uniformed network features of genes. SMOTE oversampling algorithm was adopted to balanee the posi- tive and negative samples in training SVM.The experimental results show- that the method of integrating heterogeneous network topology data ean prediet human essential genes more effeetively than those based the single network model.

关 键 词：人类必要基因 异构网络 过抽样 重启动随机游走 支持向量机 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]