Preprint introduces MetaboXcan framework to link genetically predicted metabolites with disease-associated loci

Researchers have posted a preprint to bioRxiv describing MetaboXcan, a multiomic framework that predicts plasma metabolite levels from genetic data and tests those predicted levels for association with complex traits. The approach is modelled on transcriptome-wide association studies (TWAS), which use genetic predictors of gene expression to identify disease-relevant genes at GWAS loci. MetaboXcan extends this strategy to the metabolome, a layer of molecular phenotypes that has been relatively underexplored in GWAS integration efforts.

Genome-wide association studies have identified thousands of loci for complex diseases, but the biological mechanisms connecting genetic variants to phenotypes remain poorly understood at the majority of loci. Integrating intermediate molecular phenotypes — expression quantitative trait loci (eQTLs), metabolite QTLs — can help prioritise candidate mediators. MetaboXcan builds genetically regulated metabolite predictors and uses them to test for associations with disease traits, potentially identifying metabolic pathways that mediate genetic risk.

The framework complements existing tools such as PrediXcan and S-PrediXcan, which operate on the transcriptome, by adding a metabolomic dimension. Preprint validation data and code availability are described in the manuscript. This work is a preprint and has not yet completed peer review.