BALZARINI MONICA GRACIELA
Congresos y reuniones científicas
Título:
An approach to account for genotype×environment interaction in GWAS
Autor/es:
RUEDA CALDERÓN, A.; BRUNO, C.; BALZARINI, M.
Reunión:
Conferencia; XXIXth International Biometric Conference; 2018
Resumen:
In multi-environmental agricultural trials, it is crucial to estimate the contribution of genotype (G), environment (E), and genotype×environment (G×E) effects on the variation of quantitative traits. Currently, the increasing availability of molecular marker (MM) data, allows us to estimate the G effect through an MM-driven model. Furthermore, most studies in the field of GWAS (Genome-Wide Association Study) have focused on analyzing individuals who are genetically related. Genetic relationship can be estimated with ancestry coefficients through knowledge of the pedigree or inferred from molecular marker similarities. The aim of this study was to compare statistical strategies for the estimation of G and G×E contribution to total variability in GWAS models with genetically related individuals. We worked with a public dataset consisting of 599 wheat genotypes, 1279 molecular markers, and phenotypically evaluated in 4 environments. The G×E interaction is the most important variance component. The first strategy was to estimate GWAS models by environment, considering the genetic structure through the pedigree matrix and, alternatively, through similarity of molecular marker profiles. The second strategy was to adjust a GWAS model for the whole data where G×E is incorporated using the correlation of genomic effects between environments; again the matrix of additive relationships was calculated from the pedigree as well as from molecular similarity. The Best Linear Unbiased Prediction (BLUP) of the G effects on each E was derived for each model, and G and G×E variances were estimated. The different analytic strategies resulted in similar values in terms of variance components and BLUPs of genotypes. However, the components of variance were estimated with greater precision in the multi-environmental model independently of the form used to contemplate the genetic correlation. Under abundant MM information and high G variances, the prediction of genetic merit by environment, without considering G×E, can provide genotype rankings without significant differences to that produced from multi-environmental models.