微阵列数据的多重比较

目的 介绍阳性结果错误率（FDR）及相关控制方法在微阵列数据多重比较中的应用。方法 用BH、BL、BY和ALSU四种FDR控制程序比较了3226个基因在两组乳腺癌患者中的表达差异。结果 四个程序在各自实用的范围内均将FDR控制在0．05以下，检验效能由大到小的顺序为：ALSU〉BH〉BY〉BL。ALSU程序因引入m0的估计,更为合理。不仅提高了检验效能，同时又较好地控制了假阳性错误。结论 在微阵列数据的比较中必须考虑FDR的控制，同时又要考虑提高检验效能。多重比较中，控制FDR比控制总Ⅰ型错误率（FWER）检验效能高，且更为实用。

Multiple Comparison Procedures for Microarray Data

Objective To apply false discovery rate(FDR) and its several control procedures for microarray data.Methods We demonstrate four FDR procedures,that is BH,BL,BY and ALSU,using a gene expression microarray dataset published by Hedenfalk and his colleagues and including samples from fifteen breast cancers,of which seven are from patients with known BRCA1 Mutations,eight from patients with known BRCA2 Mutations.Results The mentioned four procedures can all control the False Discovery Rate (FDR) at the desired 0.05 level under their respective conditions.The powers of the four procedures rank as ALSU>BH>BY>BL.ALSU procedure improve the power over the original procedure,mainly because it first estimate m_0 and provide tighter control of the FDR at the expense of a small proportion of erroneous rejections.Conclusion The controlling of FDR must be brought into consideration in the multiple problem for microarray data.Power should also be pay attention. And the use of FDR is a more powerful and practical approach as compared with the familywise error rate(FWER) controlling procedure.