Zhang Yuan-Ming, Lü Hai-Yan, Yao Li-Li
Section on Statistical Genomics, State Key Laboratory of Crop Genetics and Germplasm Enhancement, College of Agriculture, Nanjing Agricultural University, 1 Weigang Road, Nanjing 210095, China.
Theor Appl Genet. 2008 Sep;117(5):683-90. doi: 10.1007/s00122-008-0809-0. Epub 2008 Jun 19.
The Haseman-Elston (HE) regression, developed in the 1970s, remains in common use to detect genetic linkage between a quantitative trait and a genetic marker. Although the technique has been improved in a number of ways, it predicts a high rate of false positive quantitative trait locus (QTL) because it is based on a single-QTL model. We have extended the origin HE regression to multi-QTL HE (MQHE) regression, so that all markers across the entire genome can be exploited simultaneously. The parameters have been estimated by the penalized maximum likelihood method, and several response variables for phenotypic difference have been compared in order to optimize the procedure. The method has been tested by simulation in a pedigree population of maize inbred lines of known ancestry. These simulations show that the trait product is the optimal response variable for phenotypic difference. The false positive rate produced by the MQHE regression is substantially lower than that generated by either variance component analysis or the origin HE regression. The MQHE regression, with the trait product as the response variable, represents a significant improvement on existing methods for QTL mapping in a set of inbred lines (or cultivars) of known ancestry.
20世纪70年代开发的哈斯曼-埃尔斯顿（HE）回归，至今仍常用于检测数量性状与遗传标记之间的遗传连锁。尽管该技术在许多方面已有改进，但由于它基于单数量性状基因座（QTL）模型，所以会预测出较高的假阳性数量性状基因座率。我们已将原始的HE回归扩展为多QTL的HE（MQHE）回归，以便能同时利用整个基因组中的所有标记。参数已通过惩罚最大似然法进行估计，并且为优化该程序，已对表型差异的几个响应变量进行了比较。该方法已在已知谱系的玉米自交系家系群体中通过模拟进行了测试。这些模拟表明，性状乘积是表型差异的最佳响应变量。MQHE回归产生的假阳性率显著低于方差成分分析或原始HE回归产生的假阳性率。以性状乘积作为响应变量的MQHE回归，代表了对已知谱系的一组自交系（或品种）中QTL定位现有方法的重大改进。
