摘要
提出一种基于基因表达谱数据筛选差异表达基因的新方法;介绍了筛选差异表达基因常见方法-错误发现率方法(False Discovery Rate,FDR,),分析了多重假设检验p值性质,并根据p值性质提出了一种筛选差异表达基因新方法-单位γ度量法(Unit Measure-γ,UM-γ),建立了计算机模拟基因表达谱数据模型,制定了假阴性率、假阳性率、灵敏度、特异度以及总体错误率等作为考核指标,并使用基因表达谱模拟数据进行计算、比较;单位γ度量法估计非差异表达基因个数具有较高的稳定性和准确性;单位γ度量法既能够同时控制假阳性、假阴性以及总体错事件的发生,又能在一定程度上提高筛选结果的灵敏度和变异度;新提出的方法能有效、准确且稳定的对模拟数据差异表达基因进行筛选.
Objective To introduce a new method to screen differentially expressed genes based on gene expression data;Methods The paper introduced false discovery rate（FDR）,which is a conventional method to screen differentially expressed genes.We analyzed the multiple hypothesis testing p value properties,and put forward the new method based on p value properties.The new method named Unit Measure-gamma（UM-γ）.Gene expression data model is established with computer.This paper chose False negative Rate,False positive Rate,Specificity and Sensitivity as evaluation indexes.What＇s more,we calculated and compared different methods by the gene expression date.Results Unit Measure-gamma calculate the number of non-differentially expressed genes is more stable and accurate.Not only Unit Measure-gamma can control False positive events,false negative events and total error events,but also can improve sensitivity and Specificity.Conclusion The new method is effective,accurate and stable to screen differentially expressed genes with simulation data.
出处
《数学的实践与认识》
北大核心
2016年第18期122-128,共7页
Mathematics in Practice and Theory
基金
南京医科大学科技发展基金面上项目(NJMU20150035)
关键词
基因表达谱
差异表达基因
错误发现率
假阴性率
假阳性率
灵敏度变异度
gene expression profile date
differentially expressed genes
false discovery rate
false negative rate
false positive rate
sensitivity
degree of variation
