henhao
Sign in
Genetic & Genomic MedicineNature Genetics
Peer-reviewed journal article

A global atlas of genetic associations of 220 deep phenotypes

Saori SakaueiD, Masahiro KanaiiD, Yosuke TanigawaiD, Juha M KarjalaineniD, Mitja KurkiiD, S. KoshibaiD, Akira NaritaiD, Takahiro Konuma, Kenichi YamamotoiD, Masato AkiyamaiD, Kazuyoshi IshigakiiD, Akari SuzukiiD, Suzuki KeniD, Wataru ObaraiD, Ken YamajiiD, Kazuhisa TakahashiiD, Satoshi AsaiiD, Yasuo TakahashiiD, Takao SuzukiiD, Nobuaki Shinozaki, Hiroki YamaguchiiD, Shiro Minami, Shigeo MurayamaiD, Kozo Yoshimori, Satoshi NagayamaiD, , Yukinori OkadaiD

Abstract

Current genome-wide association studies (GWASs) do not yet capture sufficient diversity in populations and scope of phenotypes. To expand an atlas of genetic associations in non-European populations, we conducted 220 deep-phenotype GWASs (diseases, biomarkers, and medication usage) in BioBank Japan (n=179,000), by incorporating past medical history and text-mining of electronic medical records. Meta-analyses with the UK Biobank and FinnGen (ntotal=628,000) identified [~]5,000 novel loci, which improved the resolution of genomic map of human traits. This atlas elucidated landscape of pleiotropy as represented by MHC locus, where we conducted HLA fine-mapping. Finally, we performed statistical decomposition of matrices of phenome-wide summary statistics, and identified latent genetic components, which pinpointed responsible variants and biological mechanisms underlying current disease classifications across populations. The decomposed components enabled genetically-informed subtyping of similar diseases (e.g., allergic diseases). Our study suggests a potential avenue for hypothesis-free re-investigation of human diseases through genetics.

PDFDOICodeData
Full-text digest ↓

02 /Specification & Classification

Click any tag to browse matching papers
Citations
2,330
FWCI
179.45
Field percentile
top 1%
OpenAlex Topics

In-depth summary

✦ Multi-agent verified full-text synthesis

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (15 Mar 2021)

Cross-population GWAS of 220 phenotypes in BioBank Japan, UK Biobank, and FinnGen (n=628,000) identifies 5,343 previously unreported loci and decomposes summary statistics into latent genetic components for disease subtyping.

Previous genome-wide association studies have been predominantly conducted in European cohorts on pre-selected target diseases and lacked systematic frameworks to interpret shared genetic architecture across the phenome. Incorporating electronic medical records, past medical history, and prescription records in BioBank Japan alongside harmonized European biobank meta-analyses and matrix decomposition expands non-European discovery and links latent genetic components to biological pathways.

Key results

  1. ShownDeep-phenotype GWASs of 220 traits in BioBank Japan identified 519 disease, 2,249 biomarker, and 215 medication genome-wide significant loci with high effect-size concordance with European cohorts.

    Across 178,726 BBJ participants (159 diseases, 38 biomarkers, 23 medication usages; 108 traits never before studied by GWAS in East Asians), GWAS identified 519 disease, 2,249 biomarker, and 215 medication loci (P < 5.0×10^-8, including 113 disease and 281 biomarker loci not previously reported), with 94.2% (2,171 of 2,305, P < 10^-325) replicating direction in Europeans and a median genetic correlation ρge = 0.82. · Introduction; GWAS of 220 traits in BBJ and cross-population…; Genome-wide association study of 220 traits in…

  2. ShownCross-population meta-analyses across BioBank Japan, UK Biobank, and FinnGen uncovered 5,343 previously unreported genome-wide significant trait associations.

    Meta-analyzing up to 628,000 individuals yielded 1,730 disease, 12,066 biomarker, and 1,018 medication-usage loci at P < 5.0×10^-8 (844, 7,309, and 500 loci at Bonferroni P < 7.6×10^-11), of which 571 disease, 4,471 biomarker, and 301 medication loci were not previously reported. · Abstract; GWAS of 220 traits in BBJ and cross-population…; Discussion; Meta-analysis, definition of significant loci…

  3. ShownRegional pleiotropy peaked at ALDH2 and MHC in Japanese and MHC and ATXN2/SH2B3 in Europeans, with highly pleiotropic loci enriched for signatures of recent positive selection.

    In BBJ, rs671 at ALDH2 (47 traits), rs2523559 at MHC (24 traits), and rs1260326 at GCKR (20 traits) had the most associations; in Europeans, rs9265949 at MHC (46 traits) and rs7310615 at ATXN2/SH2B3 (38 traits) led, and singleton density score (SDS) χ^2 fold-enrichment increased with association count in both populations. · The regional landscape of pleiotropy; Evaluation of regional pleiotropy

  4. ShownClassical HLA fine-mapping resolved 75 independent signals in BBJ and 129 in UK Biobank, pinpointing shared amino acid variants across infectious and autoimmune diseases.

    Imputing classical HLA alleles across 159 diseases and 38 biomarkers identified 75 independent signals (53 traits) in BBJ and 129 (63 traits) in UKB (P < 5.0×10^-8), including HLA-DRβ1 Ser57 in BBJ associated with pulmonary tuberculosis (OR = 1.20, P = 7.1×10^-19), Grave's disease (OR = 1.29, P = 2.6×10^-14), Hashimoto's disease (OR = 1.50, P = 9.0×10^-8), and Sjögren's syndrome (OR = 2.04, P = 7.9×10^-12). · Pleiotropic associations in HLA and ABO locus; Fine-mapping of HLA and ABO loci

  5. ShownTruncated singular-value decomposition of cross-population disease summary statistics extracted 40 latent components that recapitulated ICD10 chapters and separated type I from type IV allergic diseases.

    DeGAs TSVD on 318 GWASs (159 diseases x 2 populations) by 22,980 LD-pruned variants explained 36.7% of variance in 40 components, linking BBJ varicose veins (ncase = 474) with EUR GWAS (ncase = 22,037) via component 11 (vascular development P = 4.2x10^-7) and splitting allergies into axis-1 (components 3, 16; IgE/Th2 P = 9.9x10^-46, 2.9x10^-44) and axis-2 (components 26, 34; IL-13/IFN P = 1.6x10^-10, 5.2x10^-9). · Deconvolution of a matrix of summary statistics…; Deconvolution of a matrix of summary statistics…

Methods, data and reproducibility

Curated medical history, text-mined EHRs, ~7M prescriptions, and biomarkers in BioBank Japan (n = 178,726) to run 220 GWASs with SAIGE and BOLT-LMM, and ran inverse-variance meta-analyses (METAL) with UK Biobank (n = 361,194) and FinnGen release 3 (ntotal = 628,000). Evaluated pleiotropy and selection via SDS, fine-mapped HLA and ABO, estimated correlations with LDSC and Popcorn, and applied DeGAs TSVD.

  • BioBank Japan (BBJ) individual genotypes, past medical history, EMR text-mining, biomarkers, and prescription records · human · n = 178,726 donors, 220 phenotypes (159 diseases, 38 biomarkers), 13,530,797 SNPs · JGAD00000000123 and JGAS00000000114
  • Summary statistics of 220 GWASs across BioBank Japan, European cohorts, and cross-population meta-analyses in NBDC · human · n = 628,000 total individuals across 220 harmonized traits · hum0197
  • UK Biobank and FinnGen release 3 genotypes and harmonized clinical/biomarker phenotypes · human · n = 361,194 UKB (196 traits) and FinnGen release 3 (128 traits; ntotal = 628,000) · 47821
  • Tohoku Medical Megabank Organization (ToMMo) and European metabolite GWAS summary statistics · human · n = 206 metabolites in ToMMo (EAS) and 248 metabolites in EUR · TGA000003
Methods & tools
SAIGE, BOLT-LMM, METAL, PLINK, Popcorn, SNP2HLA, Minimac3, IMPUTE4, DeGAs, GREAT, ANNOVAR, GREP
Code
https://github.com/bulik/ldsc
Data & accessions
http://trace.ddbj.nig.ac.jp/jga/index_e.htmlhttps://pheweb.jp/https://jmorp.megabank.tohoku.ac.jp/202008/gwas/TGA000003https://www.finngen.fi/en/access_resultshttp://www.nealelab.is/uk-biobank/ukbround2announcementhttp://metabolomics.helmholtz-muenchen.de/gwas/index.php?task=download
Software & environment
SAIGE v0.37, BOLT-LMM v2.3.4, Popcorn v1.0, GREAT v4.0.4, ssimp v0.5.5, SHAPEIT2 r837, IMPUTE4 r300.3, baselineLD v2.2, Python sklearn.decomposition TruncatedSVD, Minimac3, METAL, PLINK, PLINK2, ANNOVAR, SNP2HLA, Eagle
Reproducibility gaps
Custom analysis scripts are not deposited; full ToMMo metabolite GWAS summary statistics are being prepared as a separate project.
Limitations (3)
  • Statistical fine-mapping was not conducted across all identified loci (only for the MHC region and representative loci ALDH2 and GCKR), leaving potential confounding from linkage disequilibrium tagging in regional pleiotropy counts. · stated by the authors
  • The latent components and component ordering produced by DeGAs truncated singular-value decomposition depend on the specific set of phenotypes included in the input matrix, requiring external validation before generalizing disease reclassifications. · stated by the authors
  • Liability-scale heritability was not reported because country-specific population prevalence estimates were unavailable for all 159 diseases. · stated by the authors

Who should read:Read this if you work on trans-ancestry GWAS, biobank phenotyping, HLA fine-mapping, or matrix decomposition of phenome-wide summary statistics.

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
THISLevin '25Kanai '21Shikov '20Canela‐Xand '18DeBoever '20Chen '23
64% semantic matchGenetic & Genomic Medicine
Michael G. Levin, Satoshi Koyama, Jakob Woerner+19 more·medRxiv·Apr 18, 2025· 5 citations