Comparing GWAS models for genetically correlated multi-environment data

RUEDA CALDERÓN, A.; BRUNO, C.; BALZARINI, M.

Valencia

Conferencia; XVII Conferencia Española y VII Encuentro Iberoamericano de Biometría; 2019

Within a context of highly available molecular marker information, and with the omnipresence of multi-environment trials (MET) to assess genotype (G) performance in different environments (E), genome-wide association studies (GWAS) require models to handle multi-environment data. When the genotype (G) effects are assessed from MET, the effects of genotype by environment interaction (G×E) can impact results. Besides, prediction of the G and marker effects can be also affected by genetic correlations. The aim of this study was to compare the accuracy of different statistical approaches that use genome-wide genetic and pedigree information to account for genetic correlation in GWAS from MET. Data comprises 599 wheat lines genotyped with 1279 molecular markers assessed with 3 replicates in 4 environments. The compared models were: M1 ̶ Single-environment model fitted by adding pedigree information to account for correlations among lines; M2 ̶ Single-environment model with correlations among lines estimated from molecular similarity; M3 ̶ MET model involving pedigree information; and M4 ̶ MET model involving molecular similarity. For each model, variance components, marker scores and GBLUP were obtained. The magnitude of the genetic variability estimates depended on the information used to model genetic correlations. Selected genotypes were similar in single-and multi-environment models. However, MET models are preferred because is possible to quantify G×E and to obtain fitting criteria to select the best model for the underlying genetic correlations.