A confidence set inference procedure for gene mapping using markers with incomplete polymorphism.
A confidence set inference procedure for gene mapping using markers with incomplete polymorphism.
复制标题
使用不完全多态性标记进行基因作图的置信集推断程序。
DOI：
10.1159/000084731
复制
发表时间：
2005
期刊：
影响因子：
--
通讯作者：
Lin,Shili
中科院分区：
文献类型：
--
作者：
Papachristou,Charalampos;Lin,Shili
最近一种基于置信集推断（CSI）的基因定位方法有几个优点，包括避免多次测试的修正，具有已知统计特性的置信区间的可用性，以及疾病基因的充分定位。本文提出了一种扩展的CSI方法，可以处理不完全多态性的标记，从而提高了CSI方法集在实际情况中的适用性。仿真研究表明，新程序保留了原CSI的主要优点。虽然通常需要更多的数据来获得类似的功率，但对于杂合度为80%或更高的标记，这种增加是适度的。我们还研究了相对风险估计和疾病模型的影响。我们的分析表明，如预期的那样，来自实际相对风险或多位点疾病模型的扰动通常会导致I型误差的功率降低或膨胀。然而，对于某些类别的双位点疾病模型，CSI仍然可以表现良好，至少对一个疾病位点具有相当高的实际覆盖概率。将CSI应用于遗传分析研讨会13提供的数据产生了令人鼓舞的结果，因为它们与使用其国家物理实验室兄弟姐妹对方法从GENEHUNTER获得的数据相比有利。
A recent approach for gene mapping based on confidence set inference (CSI) promises several advantages, including avoidance of corrections for multiple tests, availability of confidence intervals with known statistical properties, and sufficient localizations of disease genes. This paper proposes an extended CSI procedure that can handle markers with incomplete polymorphism, thereby increasing the applicability of the set of CSI methods in practical situations. Simulation studies show that the new procedure retains the main advantages of the original CSI. Although it generally requires more data to achieve a similar power, this increase is moderate for markers with 80% heterozygosity or higher. We also investigate the effects of relative risk estimates and disease models. Our analyses show that perturbation from actual relative risks or multilocus disease models generally leads to reduction in power or inflation in type I error, as expected. Nevertheless, for certain classes of two-locus disease models, CSI can still perform well, with reasonably high actual coverage probabilities for at least one of the disease loci. Application of CSI to the data provided by the Genetic Analysis Workshop 13 yields encouraging results, as they compare favorably to those obtained from GENEHUNTER using its NPL sib-pair method.