| 摘要: |
| 降低和控制对大量分子标记进行测验的统计错误,已成为目前基因组研究中的普遍难题。针对此问题,本文根据序贯分析和多重假设测验的理论,介绍了可用于变化样本容量下对大量标记同时进行测验分组的序贯多重决策过程的原理和方法,探讨了该方法在实际应用中的计算精度和计算量优化等关键问题及相应的解决方案,最后以1套包括5 841个SNP和87个细胞系的药物遗传学实验数据进行实例分析,并与传统的测验结果比较,表明了序贯多重决策过程在实际应用中的优点和可行性。 |
| 关键词: 基因组,多重假设测验,统计错误,序贯分析,序贯多重决策过程 |
| DOI：10.11841/j.issn.1007-4333.2006.02.028 |
| 投稿时间：2006-03-15 |
| 基金项目:美国NIH资助项目(U01GM6334005) |
|
| Sequential multiple decision procedures and applications in genome studies |
|
|
|
| Abstract: |
| How to reduce and control statistical errors of testing larger numbers of markers has become a common and difficult problem in genome researches.To deal with this issue,under the theories of sequential analysis and multiple hypothesis test,we introduce in this paper a novel method,Sequential Multiple Decision Procedure(SMDP),which can be used for testing and grouping large numbers of makers with sequentially varying samples size.Some crucial practical aspects (computational accuracy and amount,etc.) of this method were investigated,and corresponding strategies were proposed.A set of real data from a pharmaco-genetics experiment containing 87 cell lines and 5841 SNPs was analyzed,in comparison with traditional methods,as an example to demonstrate the feasibility and advantages of SMDP. |
| Key words: genome,multiple hypothesis test,statistical error,sequential analysis,sequential multiple decision procedure(SMDP) |
