基于CFS算法研究肾癌中的关键基因

Identification of key genes in renal cell carcinoma based on CFS gene selection algorithm

目的:识别肾细胞癌(RCC)中的关键基因,并揭示其在肿瘤中的作用机理.方法:从GEO数据库下载芯片数据GSE53757,筛选它们之间的差异表达基因(DEGs).使用DAVID在线工具对DEGs进行GO功能注释和KEGG富集分析,然后使用基于特征子集相关性(CFS)的变量筛选方法筛选DEGs的关键基因,并根据筛选出的关键基因,使用支持向量机(SVM)方法建立筛选RCC样本和正常对照样本的分类预测模型.结果:共筛选到541个DEGs,包括312个上调基因和229个下调基因.选择21个作为特征基因,通过SVM方法建立RCC样本和正常对照样本之间的分类模型,其预测精度为97.2%.此外,STRING数据库筛选的Top10Hub基因中也发现了4个与CFS算法筛选出的特征基因重合的Hub基因(CD40,EGFR,CAV1和TGFA).结论:CFS是用于筛选RCC中关键基因的有用工具.并且,CD40,EGFR,CAV1和TGFA这4个基因很可能为诊断RCC的目标基因.

AIM： To identify key genes signatures in renal cell carcinoma （RCC） and uncover their potential mechanisms. METHODS： Firstly, the gene expression profiles of GSE53757, which contained 144 samples, including 72 RCC samples and 72 controls, was downloaded from GEO database. And then differen- tially expressed genes （DEGs） between the RCC samples and the controls were identified. After that, GO and KEGG enrichment analyses of DEGs were performed by DAVID. Furthermore, corre- lation-based feature subset （CFS） method was applied to the se- lection of key genes of DEGs. In addition, the classification model between the RCC samples and the controls was built by support vector machines （SVM） based on selection of key genes. RESULTS： DEGs contained 541 genes, including 312 up-regu- lated and 229 down-regulated genes. A total of 21 DEGs were selected as the feature genes to build the classification model between the RCC samples and the controls by CFS method. The accuracy of the classification model is 97.2%. Besides, four feature genes （CD40, EGFR, CAV1 and TGFA） also can been found in the top 10 hub genes screened by STRING database. CONCLUSION： It indicats that CFS is a useful tool to identify key genes in RCC. Besides, we also predicts genes such as CD40, EGFR, CAV1 and TGFA might be target genes for diagno- sing the RCC.