henhaobio
Sign in

Nucleic Acids ResearchGeneticsPeer-reviewed journal article

The NHGRI-EBI GWAS Catalog: standards for reusability, sustainability and diversity

María CerezoiD, Elliot SollisiD, Yue JiiD, Elizabeth LewisiD, Ala AbidiD, Karatuğ Ozan BircaniD, Peggy Hall, James D. HayhurstiD, Sajo JohniD, Abayomi MosakuiD, Santhi RamachandraniD, Amy ForemaniD, Arwa IbrahimiD, James Alastair McLaughliniD, Zoë May PendlingtoniD, Ray StefancsikiD, Samuel A. LambertiD, Aoife McMahoniD, Joannella MoralesiD, Thomas KeaneiD, Michael InouyeiD, Helen ParkinsoniD, Laura Wiseman HarrisiD

PDFDOIPreprintCodeData

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (23 Oct 2024) · CC BY 4.0

The GWAS Catalog now holds 625,113 curated SNP-trait associations from 6,921 publications plus >85K summary statistics datasets, with a new GWAS-SSF submission standard and GRCh38 harmonisation.

GWAS summary statistics without a common standard are hard to combine or feed into downstream tools, and the surge in molecular trait studies and sensitive population labels strains curation and reuse. This update adds a community summary statistics standard, faster harmonisation, protein and metabolite ontology mapping, a faster interface, and revised population and cohort descriptors.

Key results

  1. ShownBy mid-2024 the Catalog had grown into a very large curated human GWAS repository. Full summary statistics are available for 66% of journal-published GWAS, plus many prepublished datasets.

    625,113 SNP-trait associations (p<1×10-5) across >15.5K traits from 6,921 publications and 108,850 GWAS analyses; >85K summary statistics datasets including ~13,000 prepublished, plus 52 embargoed submissions; ~56TB stored; accessed from 205 countries by >300,000 unique users (July 2023 to June 2024). · Abstract; Introduction

  2. SuggestedGWAS-SSF, implemented in April 2023, sets mandatory and recommended columns plus YAML metadata. Monthly submissions rose after it was published, and most compliant files also include optional fields that were not required.

    33% increase in monthly submissions after publication of the standard; every GWAS-SSF submission passed mandatory data and metadata checks, with ~74% of data files and ~85% of metadata including at least one recommended field; PLINK 2.0 can now write GWAS-SSF output directly. · Improved Data content

  3. ShownA reworked harmonisation pipeline cut per-file processing time. It now provides GRCh38-aligned, forward-strand versions of nearly the whole summary statistics collection, released daily.

    Average per-file processing time reduced by 75% through pysam, Nextflow resource management and sampling 10% of variants for palindromic orientation; all GWAS-SSF files and 96% of pre-GWAS-SSF files harmonised; on average ~3.5% of variants removed as unharmonisable; outputs include a tabix index and a report file. · Improved Data content

  4. ShownMost new studies are now molecular quantitative trait GWAS (protein, metabolite, lipid). In response, the Catalog maps these traits to the Ontology of Biological Attributes with CHEBI links and redesigned the interface for very large publications.

    Molecular trait studies were 82% of studies added in 2023; across all such studies, 54% protein, 32% metabolite, 11% lipid, 3% multiple. First-half 2024 intake ~3,900 studies and ~8,600 associations per month vs 92 studies and ~3,300 associations in 2020 (42-fold, 2.6-fold); page loads up to 95% faster. · Improved Data content; UI improvements; Figure legends

  5. ShownUK Biobank is the most frequently used cohort. Its analyses are mostly limited to samples labelled European, which adds substantially to the European skew of Catalog content.

    UKB appears in 13% of publications and 30,800 GWAS, ahead of FinnGen (~2% since 2020) and BioBank Japan (~1.3%); 71% of publications and 83% of studies using UKB include only samples with the European ancestry label; >250 distinct population terms curated. · Population descriptors; Figure legends

Methods, data and reproducibility

A database update. GWAS publications are triaged with the LitSuggest machine-learning classifier, then curated by hand with EFO trait annotation and population descriptor annotation. Deposited summary statistics are checked against the GWAS-SSF standard and harmonised to GRCh38 with a Nextflow pipeline.

  • Curated SNP-trait associations from published human GWAS (snapshot as of 1st July 2024) · human · n = 625,113 associations; 6,921 publications; 108,850 GWAS
  • Full genome-wide summary statistics datasets, both journal-published and prepublished · human · n = >85K datasets (~13,000 prepublished)
  • Molecular quantitative trait GWAS (protein, metabolite, lipid measurements) added to the Catalog · human
  • Gene-environment (GxE) interaction studies · human · n = >850 studies; >10K associations
  • Studies based solely on whole genome or exome sequencing (seqGWAS) · human · n = 16K studies
  • UK Biobank-based GWAS in the cohort usage and ancestry label analysis · human · n = 30,800 GWAS
Methods & tools
LitSuggest, Experimental Factor Ontology (EFO), Ontology of Biological Attributes (OBA) with CHEBI mappings, GWAS-SSF summary statistics standard, SSF-morph web interface, pysam, Nextflow, tabix, PLINK 2.0, REGENIE, BOLT-LMM, SNPtest, SAIGE (pre-built conversion configs), GWAS Diversity Monitor, PGS Catalog
Software & environment
The harmonisation pipeline uses Python (pysam) and Nextflow and is described as portable to users' own environments. Code is under the Apache version 2.0 licence. No package versions or containers are stated.
Reproducibility gaps
Summary statistics submitted before March 2021 fall under EBI terms of use rather than CC0, so licences must be checked per dataset. 52 submissions remain embargoed. Legacy unformatted, CNV and gene-based files are not harmonised.
Limitations (6)
  • A pilot extraction found that most GWAS do not report the reference panels used to define genetic similarity, so the Catalog cannot yet record this descriptor as the 2023 NASEM report recommends. · stated by the authors
  • Catalog content stays skewed toward samples labelled European. UK Biobank analyses usually keep only White or White British participants, and ~30K minority-group individuals out of ~500K are frequently excluded. · stated by the authors
  • Mapping protein and metabolite traits to chemistry ontologies is hard at scale because metabolomics platforms differ and identifiers are ambiguous. Accurate mapping depends on study authors supplying CHEBI, Protein Ontology or UniProt IDs. · stated by the authors
  • Cohort names are inconsistent across sources (one cohort appears as MyCode, Geisinger or GHS). There is no consensus on how to describe sample environments, and no standard yet exists for reporting CNV/SV associations. · stated by the authors
  • The impact figures (33% more monthly submissions after GWAS-SSF, 80% more citations for papers that share data) come from uncontrolled before/after or observational comparisons. They are not adjusted for overall growth in GWAS output. · noted in this summary
  • The harmonisation speed-up and its correctness are documented only in supplementary figures. The main text does not quantify how accurate it is to infer palindromic variant orientation from a 10% sample instead of the full file. · noted in this summary

Read this if you reuse GWAS Catalog associations or summary statistics, deposit GWAS results, or need the Catalog's GWAS-SSF, harmonisation, cohort and population descriptor conventions.

Citations
434
FWCI
114.45
Field percentile
top 1%

Abstract

The NHGRI-EBI GWAS Catalog serves as a vital resource for the genetic research community, providing access to the most comprehensive database of human GWAS results. Currently, it contains close to 7,000 publications for more than 15,000 traits, from which more than 625,000 lead associations have been curated. Additionally, 85,000 full genome-wide summary statistics datasets - containing association data for all variants in the analysis - are available for downstream analyses such as meta-analysis, fine-mapping, Mendelian randomisation or development of polygenic risk scores. As a centralised repository for GWAS results, the GWAS Catalog sets and implements standards for data submission and harmonisation, and encourages the use of consistent descriptors for traits, samples and methodologies. We share processes and vocabulary with the PGS Catalog, improving interoperability for a growing user group. Here, we describe the latest changes in data content, improvements in our user interface, and the implementation of the GWAS-SSF standard format for summary statistics. We address the challenges of handling the rapid increase in large-scale molecular quantitative trait GWAS and the need for sensitivity in the use of population and cohort descriptors while maintaining data interoperability and reusability.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISHayhurst '22Elsworth '20Reales '22Morales '18Lyon '21Thelwall '19
70% semantic matchBioinformatics
James D. Hayhurst, Annalisa Buniello, Laura Wiseman Harris+20 more·bioRxiv·Jul 15, 2022· 16 citations