Skip to content

OmicsPred

Summary

OmicsPred is a centralised, web-accessible repository for the deposition, dissemination, and genetic prediction of multi-omic traits modeled on FAIR data principles. It unifies molecular imputation models across transcriptomic, proteomic, and metabolomic traits, supporting downstream transcriptome-, proteome-, and metabolome-wide association studies (TWAS, PWAS, and MWAS). As of May 2026, the platform hosts 3,339,469 prediction models trained on diverse cohorts and genetic ancestries.

Key Features and Datasets

OmicsPred harmonizes prediction models from various datasets and distributes them in formats compatible with PGS Catalog Calculator and MetaXcan: - PredictDB Models: Hosts transcriptomic and splicing models from GTEx V8 (Elastic Net and MASHR) across 49 human tissues. - INTERVAL Cohort: The INTERVAL BioResource contributes 17,227 models predicting blood transcriptomics, proteomics (profiled via Olink and SomaLogic), and metabolomics (profiled via Nightingale NMR and Metabolon MS). - Diverse Ancestries: Features SomaLogic-based plasma proteomics prediction models from the TOPMed Multi-omics pilot study (3,578 models) and the ARIC study (2,733 models). - UK Biobank (UKB): Includes Olink proteomics models trained in European ancestry (2,612 models) and multi-ancestry (2,704 models) subsets. - Phenome-wide Insights: Integrates phenome-wide association study (PheWAS) results utilizing the Million Veterans Program (MVP) and UK Biobank, mapping traits to EFO ontologies and Reactome pathways.

  • AGER — a cross-ancestry PheWAS signal highlighted by OmicsPred models.
  • Coefficient of Determination (R-squared) — a core measure of molecular-trait prediction performance.
  • Inouye Lab — contributor to the platform and its systems-genomics applications.
  • Lambert Lab — contributor to model standardization and dissemination.

Citations

  • Foguet, C., Gil, L., Xu, Y., Salazar-Magaña, S., Ritchie, S. C., Persyn, E., Im, H. K., Inouye, M., & Lambert, S. A. (2026). OmicsPred as a centralised resource for genetic prediction of multi-omic traits. medRxiv preprint. DOI: 10.64898/2026.05.15.26353298. Source paper: 2026.05.15.26353298v2.full.pdf