On the parsing of statistical information in family-based association testing.
On the parsing of statistical information in family-based association testing.
复制标题
基于家庭的关联测试中统计信息的解析。
DOI：
10.1038/ng0307-281
复制
发表时间：
2007
期刊：
影响因子：
30.8
通讯作者：
Lange,Christoph
中科院分区：
文献类型：
--
作者：
McQueen,MatthewB;Weiss,Scott;Laird,NanM;Lange,Christoph
自然遗传学小的等位基因频率很难在独立的研究中复制。在这种情况下，这些真实的发现将被视为假阳性。纯P值驱动的分析可能导致在大型全基因组扫描中选择具有相对低等位基因频率或小效应大小的snp。一种最大限度地检测“可复制”snp的分析策略是可取的，并且结合等位基因频率和遗传效应大小可以提高复制的机会。使用估计的统计能力作为直观的尺度，Van Steen方法筛选了5 - 8个snp，这些snp具有足够的效应大小和等位基因频率，可以在同一数据集中进行测试，并在独立研究中进行复制。功率计算表明，对于数以千计的样本，合理的等位基因频率和效应大小可以获得阳性结果。然而，在现实中，这种规模的绝大多数研究将包括通过多个不同的来源群体确定的样本，这些群体可能在表型和遗传背景方面存在差异。在这些情况下（特别是在不考虑效应大小的情况下），最小的P值也可能
282 VOLUME 39| NUMBER 3| MARCH 2007| NATURE GENETICS small allele frequencies will be difficult to replicate in independent studies. In such cases, these true findings will be considered as false positives. A purely P value–driven analysis may result in the selection of SNPs with relatively low allele frequency or small effect size in large genome-wide scans. An analysis strategy that maximizes the power to detect ‘replicable’SNPs is desirable, and the incorporation of allele frequency and genetic effect size enhances the chance of replication. Using the estimated statistical power as an intuitive yardstick, the Van Steen approach5–8 screens for SNPs that have sufficient effect sizes and allele frequencies to be tested in the same data set and replicated in independent studies.Power calculations suggest that positive results can be obtained for reasonable allele frequencies and effect sizes for samples numbering in the thousands. In reality, however, the vast majority of studies of this magnitude will consist of samples ascertained via multiple dissimilar source populations that may differ with respect to phenotypic and genetic backgrounds9. Under these circumstances (in particular, when effect sizes are not taken into consideration) the smallest P values may also