【作者】 曾平； 黄水平； 赵华硕； 金英良； 谭英；

【Author】 Zeng Ping,Huang Shuiping,Zhao Huashuo,Jin Yingliang,Tan Ting Department of Epidemiology and Health Statistics,XuZhou Medical College(221002),Xuzhou

【机构】 徐州医学院流行病与卫生统计学教研室；

【摘要】 目的研究基于错误发现率的微阵列数据多重假设检验和估计。方法采permutation检验获得原始P值,多重检验中控制和估计FDR,并应用于Hedenfalk的乳腺癌微阵列。结果如选择|t|≥3作为拒绝域,则识别出244个表达差异的基因,对应的FDR估计值大约为8.5%。结论 permutation检验适合于微阵列数据,通过考虑零假设比例的情况能够进一步提高FDR控制的效率和灵活性。更多还原

【Abstract】 Objective To study the multiple hypotheses testing in microarray data based on permutation resampling and estimate the false discovery rate(FDR).Methods The original p values were obtained using permutation test,then control and estimate FDR in multiple hypotheses testing,Hedenfalk’s microarray data about breast cancer was illustrated as example.Results 244 genes were identified as differential expression if the given rejection area was specified as \t\ > 3 with an estimated FDR of 8.5%. Conclusion Permutation test was appropriate for microarray data analysis,the effectiveness and flexibility of false discovery rate could be improved by taking the information of the proportion of null hypotheses into account.更多还原