Comparison of GWA statistical methods for traits under different genetic structures: A simulation study
Nenhuma Miniatura disponível
Data
2020-11-01
Autores
Garcia, Baltasar Fernandes [UNESP]
Melo, Thaise Pinto de [UNESP]
Neves, Haroldo Henrique de Rezende
Carvalheiro, Roberto [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Resumo
Several methods have been used to perform genome wide association (GWA) studies, aiming to map quantitative trait loci (QTL) and candidate variants. The single-step genomic best linear unbiased prediction (ssGBLUP) method is one alternative to perform GWA, which allows the simultaneous use of genotypic, pedigree and phenotypic information. Bayesian multiple regression models have also been used to perform GWA, allowing specifying different priori distribution for the molecular marker effects. The aim of the present study was to evaluate, through simulation, the performance of different methods in the identification of QTLs for polygenic and major gene traits, under different heritabilities when only relatively few animals were genotyped. We also investigated the consequence of considering the agreement among results from different methods as a strategy to decrease errors associated with false positives in QTL mapping. For polygenic scenarios, results showed low power to detect QTL, irrespective of the method, and the use of phenotypes from non-genotyped animals slightly helped QTL detection of ssGBLUP and weighted ssGBLUP (wssGBLUP). For scenarios with major gene effects, there was greater power in QTL detection, with a slight superiority of Bayes C over the other methods. The inclusion of additional phenotypic information (from non-genotyped animals) harmed the performance of wssGBLUP in the presence of major QTL. When only consensus regions identified by different methods were considered as evidence of QTL, the percentage of top windows containing a true QTL tended to increase with the number of methods that identified a top window, for all scenarios. However, an important proportion of true QTL were identified only by a single or few methods. Despite the small differences among methods in the QTL detection, for polygenic traits, the ssGBLUP and wssGBLUP methods seemed to show better results in comparison to the other methods mainly in the low heritability scenario. For traits with major gene effects, the Bayes C method is expected to present better results, compared to the other methods evaluated, given a limited number of genotyped animals. The strategy to use agreement of GWA results among methods to map QTL helps to reduce false positives but comes at a cost of missing important QTL, irrespective of the genetic structure of the trait.
Descrição
Palavras-chave
Bayesian methods, GWA, QTL mapping, weighted single-step GBLUP
Como citar
Livestock Science, v. 241.