利用基因表达谱提取乳腺癌细胞分化相关特征基因  

Feature Genes Selection for Histologic Grading of Cancer

作　　者：叶云[1,2] 王桂平[1,3] 杨晓勤[1] 梁爽[1] 郑文岭[1] 马文丽[1] 

机构地区：[1]南方医科大学基因工程研究所,广州510515 [2]广西工学院生物与化学工程系,柳州545006 [3]广州医学院护理学院药理学教研室,广州510182

出　　处：《热带医学杂志》2010年第4期377-379,393,共4页Journal of Tropical Medicine

基　　金：广东省生物芯片重点实验室基金(No.2004B60144)

摘　　要：目的利用基因芯片数据挖掘识别与乳腺癌组织学分级相关的特征基因,对乳腺癌的临床诊断和生物医学研究起到借鉴和参考作用。方法从公共基因芯片数据库GEO(gene expression omnibus)获得乳腺癌芯片表达数据,利用支持向量机提取获得不同组织学分级的肿瘤样本的特征基因,并对这些基因进行生物学功能分析。结果获得了64个特征基因,分类正确率达到100%,这些基因与癌症有较大的相关性,主要集中在转录调控、离子运输、器官发生发育等多个生物学途径中。结论通过对基因芯片数据的挖掘,可以从全局上了解肿瘤的表达情况,加深对乳腺癌细胞分化分子机制的认识。Objective To identify the genes associated with histologic grading of breast cancer. Methods DNA microarray data mining method was used in this study.Gene expression profile data were obtained from the GEO database. Discriminative genes were identified by the support vector machines and then were annotated. Results 64 feature genes were selected and the accuracy was 100% .Most of these genes are associated with cancer and the functions are mainly involved in ion transport, morphogenesis and regulation of transcription. Conclusion Data mining by bioinformatics may provide a better insight into the molecular characteristics of the histologic grading of breast cancer.

关 键 词：表达谱 乳腺癌 组织学分级 支持向量机 特征基因 

分 类 号：R737.9[医药卫生—肿瘤]