A Linear Mixed Model for Joint Multi-QTL Mapping of the Body Composition Traits of Chickens

  • Jinhua Ye


 In domestic animals, resource populations used for mapping quantitative trait loci (QTLs) invariably comprise multiple families, none of which is a strictly genetically designed population. Thus, in addition to multiple QTLs, the phenotypic effects of polygenes related to multiple family pedigrees should be considered in QTL mapping. Here, we propose a mixed model containing both major genes and minor polygenes. Initially, the LASSO compression technology was used to estimate non-zero major gene loci in a linear mixed model (LMM), and then fixed-effect QTLs and the random polygenic genetic effects predicted by restricted maximum likelihood were simultaneously embedded into the LMM. Finally, we performed QTL genetic parameter estimation and linkage analysis of gene mapping. Simulation studies showed that our proposed method has advantages compared with existing LASSO and random model methods regarding statistical power to detect QTLs and estimate genetic parameters. Our method not only inherits the computational speed of LASSO but is more robust in terms of genetic parameter estimation than LMM. Real data analyses for two chicken F2 populations illustrated the application of the mapping method. By replacing the pedigree relationship matrix with a marker-based matrix, the new method can be directly applied to solve the LMM for genome-wide association studies in humans and other species.