Controlling false positives in the mapping of epistatic QTL
Controlling false positives in the mapping of epistatic QTL
复制标题
DOI：
10.1038/hdy.2009.129
复制
发表时间：
2010-04-01
期刊：
影响因子：
3.8
通讯作者：
de Koning, D-J
中科院分区：
文献类型：
--
作者：
Wei, W-H;Knott, S.;de Koning, D-J
本研究解决了在两两上位性数量性状基因座（QTL）定位中的假阳性率（FPR）控制问题。开发了一个嵌套测试框架，以（1）允许直接使用预先识别的QTL来检测一维基因组扫描中的上位性，（2）检测二维基因组扫描中的新型上位性QTL对，以及（3）推导出全基因组阈值通过排列并处理多次测试。我们使用大规模模拟来评估一维和二维方法在映射不同形式和水平的上位性方面的性能，并生成FPR，功率和准确性的配置文件，以通知上位性映射研究。我们表明，嵌套测试框架和全基因组阈值是必不可少的控制FPR在5%的水平。一维方法在检测QTL相关上位性方面通常比二维方法更强大，并且几乎识别了从二维方法检测到的所有上位性对。然而，只有二维方法可以检测上位性QTL的弱主效应。结合这两种方法可以有效地映射不同形式的上位性，而使用嵌套测试框架保持FPR在控制之下。这种方法为高通量上位性分析提供了一个很好的搜索引擎。Heredity（2010）104，401-409; doi：10.1038/hdy.2009.129; 2009年9月30日在线发表
This study addresses the poorly explored issue of the control of false positive rate (FPR) in the mapping of pair-wise epistatic quantitative trait loci (QTL). A nested test framework was developed to (1) allow pre-identified QTL to be used directly to detect epistasis in one-dimensional genome scans, (2) to detect novel epistatic QTL pairs in two-dimensional genome scans and (3) to derive genome-wide thresholds through permutation and handle multiple testing. We used large-scale simulations to evaluate the performance of both the one-and two-dimensional approaches in mapping different forms and levels of epistasis and to generate profiles of FPR, power and accuracy to inform epistasis mapping studies. We showed that the nested test framework and genome-wide thresholds were essential to control FPR at the 5% level. The one-dimensional approach was generally more powerful than the two-dimensional approach in detecting QTL-associated epistasis and identified nearly all epistatic pairs detected from the two-dimensional approach. However, only the two-dimensional approach could detect epistatic QTL with weak main effects. Combining the two approaches allowed effective mapping of different forms of epistasis, whereas using the nested test framework kept the FPR under control. This approach provides a good search engine for high-throughput epistasis analyses. Heredity (2010) 104, 401-409; doi: 10.1038/hdy.2009.129; published online 30 September 2009