Incorporating LASSO effects into a mixed model for quantitative trait loci detection

被引：16

作者：

Foster, Scott D.

Verbyla, Arunas P.

Pitchford, Wayne S.

机构：

[1] Univ Adelaide, Sch Agr Food & Wine, Glen Osmond, SA 5064, Australia

[2] CSIRO Math & Informat Sci, Hobart, Tas 7001, Australia

来源：

JOURNAL OF AGRICULTURAL BIOLOGICAL AND ENVIRONMENTAL STATISTICS | 2007年 / 12卷 / 02期

关键词：

adjusted scores; partial laplace approximation; quantitative trail loci; restricted likelihood; subset selection;

D O I：

10.1198/108571107X200396

中图分类号：

Q [生物科学];

学科分类号：

07 ; 0710 ; 09 ;

摘要：

The identification of quantitative trait loci (QTL) can be viewed as a subset selection problem. In a simulation study the least absolute selection and shrinkage operator (LASSO) is shown to be a useful and powerful tool for QTL identification. LASSO effects are embedded into a mixed model allowing simultaneous modeling of genetic and experimental effects. This provides the flexibility to model the experiment in conjunction with the power of LASSO QTL identification. Estimation is performed using an approximation to the restricted likelihood and modified Gaussian elimination. The extended mixed model is used to analyze a cattle gene mapping dataset.

引用

页码：300 / 314

页数：15

共 30 条

[1] Genetic variation in growth and body dimensions of Jersey and Limousin cross cattle. 1. Pre-weaning performance [J].