Amos C I, Krushkal J, Thiel T J, Young A, Zhu D K, Boerwinkle E, de Andrade M
Department of Epidemiology, University of Texas M.D. Anderson Cancer Center, Houston 77005, USA.
Genet Epidemiol. 1997;14(6):743-8. doi: 10.1002/(SICI)1098-2272(1997)14:6<743::AID-GEPI30>3.0.CO;2-O.
We compared several strategies for identifying and estimating effects from a genetic locus in the etiology of a complex trait. For our analyses we used data from simulated trait 1 and chromosome 5. Results from analysis of the first 20 replicates showed that a components of variance test provided considerably better power for identifying linkage than tests that consider pair differences. We also compared the power from constructing tests with a single marker, an approximate method using five markers jointly, or a multipoint analysis using all 25 markers on chromosome 5 jointly. Results from this analysis showed substantially better power when all markers were jointly used in the analysis. Results from considering all replicates showed that all methods of estimation provided maximal test statistics at the correct marker position, but the components of variance procedure provided more power to detect the correct position than other methods.
我们比较了几种用于识别和估计复杂性状病因中基因座效应的策略。在分析中，我们使用了来自模拟性状1和5号染色体的数据。对前20次重复分析的结果表明，方差成分检验在识别连锁方面的效能比考虑配对差异的检验要高得多。我们还比较了使用单个标记构建检验、联合使用五个标记的近似方法或联合使用5号染色体上所有25个标记进行多点分析的效能。该分析结果表明，在分析中联合使用所有标记时效能显著更高。考虑所有重复的结果表明，所有估计方法在正确的标记位置都提供了最大检验统计量，但方差成分程序比其他方法更有能力检测到正确位置。
