Precise Conditions for Permutation Tests to Control Multiple Testing Error Rates

排列测试的精确条件以控制多个测试错误率

• 批准号: 1007794
• 负责人: Jason Hsu
• 金额: $ 8.64万
• 依托单位: Ohio State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2010
• 资助国家: 美国
• 起止时间: 2010-08-01 至 2013-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1007794&HistoricalAwards=false
• 关键词: Precise; Conditions; Permutation; Tests; Control

The purpose of this project is to improve the control of multipletesting error rates. A popular approach to multiple testing is based onpermutation. One motivation for the use of permutation-basedtesting is to reduce the conservativeness of the tests while stillcontrolling error rates. Permutation-based tests are thought to achievethis by incorporating the multivariate distribution of the multiple teststatistics. However, the hypotheses for which permutation testing is knownto be appropriate are different from those usually considered in multipletesting. In this project, the investigators precisely describe modelassumptions and conditions for permutation testing to control multipletesting error rates. In particular, the investigators study precise modelassumptions connecting the marginal distributions of interest in multipletesting to the joint distributions known to be appropriate for permutationtesting.While the methods studied in this project can be used in many fields, theywill be particularly applicable to genetic association studies.Genome-Wide Association Studies (GWAS) are conducted to discoverbiological pathways and prognostic biomarkers, and to guide treatment.Some GWAS discoveries have failed to replicate, possibly due tostatistical testing procedures lacking error rate control. In thisproject, the investigators study the conditions under which currently usedmethods are guaranteed to appropriately control statistical error rates.Such control will lead to better return on investments in developing drugsand treatments.

该项目的目的是提高多重测试错误率的控制。 多重测试的流行方法是基于排列。使用基于排列的测试的动机之一是降低测试的保守性，同时仍然控制错误率。 基于排列的检验被认为可以通过合并多个检验统计量的多元分布来实现这一点。然而，已知适合排列检验的假设与多重检验中通常考虑的假设不同。在这个项目中，研究人员精确描述了排列测试的模型假设和条件，以控制多重测试错误率。 特别是，研究人员研究了精确的模型假设，将多重测试中感兴趣的边际分布与已知适合排列测试的联合分布联系起来。虽然本项目中研究的方法可用于许多领域，但它们特别适用于遗传关联研究。进行全基因组关联研究 (GWAS) 的目的是发现生物学途径和预后生物标志物，并指导治疗。一些 GWAS 发现未能复制，可能是由于统计测试程序所致缺乏错误率控制。 在该项目中，研究人员研究了当前使用的方法能够保证适当控制统计错误率的条件。这种控制将带来更好的药物开发和治疗方法的投资回报。