Exposome-Wide Association Study

The analytic logic is deliberately the same, but the variables differ. A GWAS scans fixed genetic variants for association with a trait; an ExWAS scans measured environmental exposures — biomarkers, chemicals, diet, and behaviors — against an outcome. Both fit one model per variable and apply stringent multiple-testing correction and replication. The crucial contrast is that the genome is fixed and measured almost without error, whereas the exposome is dynamic, correlated, and noisily measured, which makes confounding, reverse causation, and exposure measurement error far more serious concerns in ExWAS than in GWAS.

No. ExWAS is a discovery and prioritization tool that produces associations under covariate adjustment, not causal estimates. Because environmental exposures cluster together and can proxy for unmeasured confounders or for reverse causation, a surviving, replicated hit should be read as a strong, prioritized hypothesis. Establishing causality requires follow-up: triangulation with other designs, sensitivity analyses such as E-values for unmeasured confounding, mechanistic evidence, or instrumental approaches. ExWAS narrows the field of candidate exposures so that scarce causal-inference effort is spent on the most promising leads.

By borrowing the GWAS discipline. Because the scan runs one test per exposure across a large panel, the analysis pre-specifies a stringent error-control rule — typically a Bonferroni threshold of alpha divided by the number of exposures, or a false-discovery-rate procedure that bounds the expected fraction of false positives among the hits. Only associations passing this threshold are carried to replication in an independent sample. This two-stage filter is what makes the breadth of the search statistically honest and distinguishes ExWAS from simply running hundreds of separate tests and reporting whatever reaches nominal significance.