基于相对风险的肿瘤基因差异表达分析

目的寻找与肿瘤相关的基因诊疗中差异表达基因提取的方法。方法将基因表达谱数据进行预处理,采用相对风险方法筛选出差异表达基因特征子集,计算其样本间距离,然后对特征基因加权排序和过滤冗余基因,最后应用分类器对卵巢癌基因数据集进行分析,测试该方法的有效性。结果选取20维特征基因,进行分类测试,当特征基因为3～5、7和12～20维时,分类准确率可以达到100%,假阳性率可以达到0,表现出较好的可靠性,能够有效地将2个样本类型分开。结论经分类器测试证明,分类精度高,效果优于使用传统的基因差异表达分析方法。

Gene differential expression analysis for tumor based on relative risk

Objective To study the differentially expressed gene extraction method for diagnosis and treatment genes associated with tumor.Methods The gene expression spectrum data were preprocessed,and the differentially expressed genes feature subset was screened by relative risk method.The distances in classification and classification of each other were calculated,and the feature genes were weighed and ranked.Finally redundant genes were filtered in terms of its redundant coefficient between genes,the effectiveness of ovarian cancer gene data were analyzed and tested with classifier.Results Dimensions of 20 feature genes were selected to compare and analyze,the rate of accuracy reached at 100 % for false positive rate reached at 0 when feature genes of 3-5,7 and 12-20 dimensions,which was showed good reliability,and selected 2 types of feature genes effectively.Conclusion It is demonstrated that the classification accuracy is improved in comparing with conventional classification methods of differential expressed genes.