拓扑推断乳腺癌的活性miRNA介导的子通路  

Topological Inference of the Active miRNA Mediated Subpathways of Breast Cancer

作　　者：宁梓淯[1] 张建[1] 关梦 周新源 Ning Ziyu;Zhang Jian;Guan Meng;Zhou Xinyuan(Department of Medical Informatics,Harbin Medical University(Daqing),Daqing Heilongjiang 163319,China)

机构地区：[1]哈尔滨医科大学大庆校区医学信息学系,黑龙江大庆163319

出　　处：《信息与电脑》2018年第4期41-42,46,共3页Information & Computer

基　　金：大庆市指导性科技计划项目(项目编号:zd-2016-141)

摘　　要：笔者获取了人类通路数据、TCGA(The Cancer Genome Atlas)中乳腺癌表达数据、mirTarBase和miRBase等数据库中用于实验验证的miRNA-mRNA靶向关系数据,通过整合生物网络的拓扑信息、miRNAs和基因表达的差异性以及miRNAs和基因的靶向互做关系,计算出每个miRNA介导的自通路的活性值,再从中筛选出活性较高的自通路作为癌症分子诊断的标志物,进而提出一种算法miDRW。实验证明,乳腺癌的数据内5倍交叉验证的AUC值可达0.994 8,准确率达98.86%,明显高于单基因和单miRNA标志物的分类指标。可见,笔者提出的整合算法可以精确识别乳腺癌的诊断标志,并实现精确分类。The most common malignant cancers in women is breast cancer,which is complex pathogenesis and highly heterogeneous.Thus,it can’t obtain the accurate clinical molecular diagnosis with the low-throughout experiment.This paper obtains the pathway information,the breast Cancer expression data of TCGA(The Cancer Genome Atlas),the target relationship between miRNAs and gene of mirTarBase and miRBase database,and then integrates the topology information,the differences of miRNAs and gene expression,as well as the relationship between miRNAs and gene,calculates the activity of each miRNA-mediated subpathway.Finally,the high activity subpathways are called breast cancer biomarkers.The results showed that the AUC value of 5 times crossvalidation of breast cancer was 0.994 8,and the accuracy rate was 98.86%,which was significantly higher than that of individual gene and miRNA marker.It can be seen that the integration strategy can accurately identify the biomarkers of breast cancer and achieve accurate classification.

关 键 词：癌症 分子标志物 miRNA介导的子通路 

分 类 号：R737.9[医药卫生—肿瘤]