Skip to main navigation Skip to search Skip to main content

MetaboXcan: A multiomic framework linking genetically predicted metabolites, gene expression, and complex traits

  • Festus Nyasimi
  • , Sarah Sumner
  • , Yanyu Liang
  • , Xianyong Yin
  • , Yoson Park
  • , Andrew Brown
  • , Sulaiman A. Lawal
  • , Ana Viñuela
  • , Lilian Fernandes Silva
  • , Chunyu Liu
  • , Sarah D. Slack
  • , Erika Esquinca
  • , Randi K. Johnson
  • , Katerina Kechris
  • , Markku Laakso
  • , Michael Boehnke
  • , Eric Fauman
  • , Abraham A. Palmer
  • , Sandra Sanchez-Roige
  • , Ani Manichaikul
  • Nicholette D. Palmer, Hae Kyung Im

Research output: Working paper/Preprint › Preprint

9 Downloads (Pure)

Abstract

Genetically regulated molecular phenotypes, such as gene expression and metabolites, are widely thought to mediate the effect of disease-relevant genetic loci identified by genome-wide association studies (GWAS); however, molecular mechanisms connecting genetic variation to disease remain poorly understood. While transcriptome-wide association studies (TWAS) have identified disease-associated genes, strategies that integrate the metabolome remain underexplored. Here, we introduce MetaboXcan, a framework for predicting plasma metabolite levels from genetic data and associating them with complex traits using GWAS summary statistics. We trained lasso regression models on plasma metabolite and genotype data from the Metabolic Syndrome in Men Study (METSIM); these models outperform existing genetic metabolite predictors and generalize well across independent cohorts. MetaboXcan leverages these models to perform four complementary association analyses: (i) gene-to-trait (TWAS), (ii) gene-to-metabolite (M-TWAS), (iii) metabolite-to-trait (MWAS), and (iv) gene-expression–based metabolite-to-trait associations (g-MWAS). The framework further organizes results into metabolic pathways and gene–metabolite interaction networks to facilitate biological interpretation. Applied to chronic kidney disease (CKD), MetaboXcan identified known disease risk genes (PDILT/UMOD, SPATA5L1/GATM), as well as multiple CKD-relevant metabolites (e.g., glycine, homoarginine). Our integrated multiomic analysis enables biological interpretation by revealing glycine availability–centered biochemical axes spanning oxidative stress, cellular energetics, and vascular signaling, while nominating candidate mechanisms for downstream investigation. These results validate known disease-related biology with genetic evidence and generate new hypotheses for further investigation. MetaboXcan is publicly available and broadly applicable to any GWAS phenotype.
Original languageEnglish
PublisherBioRxiv
Number of pages50
DOIs
Publication statusPublished - 17 Jul 2026

Fingerprint

Dive into the research topics of 'MetaboXcan: A multiomic framework linking genetically predicted metabolites, gene expression, and complex traits'. Together they form a unique fingerprint.

Cite this