Summary

Genome-wide association studies (GWAS) allow the identification of genomic regions associated with phenotypic traits. In plant genetics, these GWAS help identify genetic factors influencing essential agronomic traits, such as yield, disease resistance, or tolerance to environmental stresses. However, GWAS have significant limitations, particularly in terms of statistical power, as they analyze millions of genetic markers for a  relatively small number of individuals. Moreover, standard GWAS approaches are generally limited to the study of associations with a single trait at a time, which restricts their ability to capture the complex genetic architecture underlying multiple traits. In human genetics, meta-analysis has been effectively used to integrate the results of multiple GWAS studies and overcome the limitations of standard approaches, allowing on the one hand to improve statistical power and on the other hand to study associations between multiple traits. Despite this potential, meta-analysis has been rarely used in plant genetics.

In this thesis, we present the reasons making existing meta-analysis methods inappropriate for applications in plant genetics and introduce new meta-analysis methods adapted to three different key contexts. First, we propose a meta-analysis method to study geneenvironment interactions, which play a crucial role in the genotype-phenotype relationship
in plants. Second, we present a meta-analysis approach to detect pleiotropic variants, which influence multiple phenotypic traits and allow a better understanding of genetic pathways. Third, we introduce a new strategy to detect genetic variants associated with groups of molecular traits, such as gene expressions or protein abundances. The methods developed in this thesis rely on latent variable models, allowing for flexible modelling of the genetic effects across different contexts. By addressing the main limitations of metaanalyses, this work offers new statistical methods for association studies in plant genetics, enabling more comprehensive investigations of the genetic architecture of complex traits.

Contact

• Annaig de Walsche (UMR GQE Le Moulon)
• Linkedin

Keywords: meta-analysis, latent variables models, genome-wide association study, plant
genetics

Publications

Journal articles

• Annaïg de Walsche, Alexis Vergne, Renaud Rincent, Fabrice Roux, Stéphane Nicolas, et al.. metaGE: Investigating genotype x environment interactions through GWAS meta-analysis. PLoS Genetics, 2025, 21, pp.e1011553. ⟨10.1371/journal.pgen.1011553⟩. ⟨hal-04994005⟩
• Annaïg de Walsche, Franck Gauthier, Alain Charcosset, Tristan Mary-Huard. Large-scale composite hypothesis testing for omics analyses.. 2024. ⟨hal-04508559⟩