Fig. 1: Overview of the analysis workflow.

We first obtained SNP-environment interaction effect estimates for the 25 environments (GWEISs), as well as SNP main effects (GWAS). Results from these analyses were used to perform gene and gene-set analyses. The effect of SNPs, genes, and gene-sets that reached standard genome-wide significance (i.e., not corrected for the 25 environments) in the interaction-based analyses were compared with their corresponding main effects. Interaction-based polygenic risk scores (iPRSs) were constructed in an independent subset of the sample to predict neuroticism: this was done by modelling the interactions between each environment and a polygenic risk score constructed based on either the SNP-environment interaction effect from GWEIS (iPRSGxE), as well as the SNP main effect from GWAS (iPRSG) as a comparison.