Association analysis in structured plant populations, an adaptive mixed LASSO approach
Dong Wang
Association analysis in structured plant populations, an adaptive mixed LASSO approach - 2010 - 1 page
Recently, there has been heightened interest in performing association analysis in important crop species. The development of mixed linear models for plant association mapping has significantly advanced the statistical methodology in this field. However, the mixed linear model has been mostly limited to single marker analysis. On the other hand, the lack of knowledge on epistasis and GxE interactions has become one of the major impediments of utilizing genomic information for crop improvement. We report the development of the adaptive mixed LASSO method that can incorporate a large number of predictors while simultaneously accounting for the population structure. LASSO can deal with situations where the number of explanatory variables is much larger than the sample size, which is not feasible for traditional regression methods. By extending adaptive LASSO to include random effects for structured populations, we can readily apply our method to the setting of plant association mapping. Our results show that the adaptive mixed LASSO method is very promising in modeling multiple genetic effects (main QTL effects and epistasis) as well as modeling gene by environment interactions when a large number of markers are available and the population structure cannot be ignored. Since no equivalent method has been proposed in the setting of crop association analysis, it is expected to have a significant impact on the study of complex traits in important crop species. Applications to wheat breeding programs has been planned with the potential of influencing plant breeding practices.
English
Association analysis in structured plant populations, an adaptive mixed LASSO approach - 2010 - 1 page
Recently, there has been heightened interest in performing association analysis in important crop species. The development of mixed linear models for plant association mapping has significantly advanced the statistical methodology in this field. However, the mixed linear model has been mostly limited to single marker analysis. On the other hand, the lack of knowledge on epistasis and GxE interactions has become one of the major impediments of utilizing genomic information for crop improvement. We report the development of the adaptive mixed LASSO method that can incorporate a large number of predictors while simultaneously accounting for the population structure. LASSO can deal with situations where the number of explanatory variables is much larger than the sample size, which is not feasible for traditional regression methods. By extending adaptive LASSO to include random effects for structured populations, we can readily apply our method to the setting of plant association mapping. Our results show that the adaptive mixed LASSO method is very promising in modeling multiple genetic effects (main QTL effects and epistasis) as well as modeling gene by environment interactions when a large number of markers are available and the population structure cannot be ignored. Since no equivalent method has been proposed in the setting of crop association analysis, it is expected to have a significant impact on the study of complex traits in important crop species. Applications to wheat breeding programs has been planned with the potential of influencing plant breeding practices.
English