GWASAD GWAS
Alzheimer's disease GWAS: Bellenguez et al. 2022
Get the data
Controlled access. Request data access on the AD Knowledge Portal; you need a free account there. See How to get the data.
| Where | ID | Access | What you get |
|---|---|---|---|
| AD Knowledge Portal | syn69670625 | Controlled | Files described below |
pip install synapseclient
synapse login
synapse get -r syn69670625Every AD Knowledge Portal folder that mentions this dataset 13 locations
Variant and gene summaries2
- AD_loci_summary › AD_finemapping · syn69865805
AD_Bellenguez_2022.gwas.top_loci.bed.gzAD_Bellenguez_EADB_2022.gwas.top_loci.bed.gzAD_Bellenguez_EADI_2022.gwas.top_loci.bed.gzAD_Bellenguez_main_2026.gwas.top_loci.bed.gzAD_Bellenguez_noproxy_2026.gwas.top_loci.bed.gzAD_Bellenguez_nobiobank_2026.gwas.top_loci.bed.gz
- AD_loci_summary › AD_xQTL_colocalization · syn69865816
ROSMAP_tpQTL_K5_AD_Bellenguez_2022.colocalization.txt.gzROSMAP_tpQTL_K5_AD_Bellenguez_2022.ADgenes_colocalization.txt.gzROSMAP_tpQTL_K3_AD_Bellenguez_2022.colocboost_sorted.bed.gz
xQTL and AD GWAS models11
- AD_finemapping › susie_rss › batch{1..2} · syn77238945
AD_Bellenguez_2022.*.gwas_finemap.rdsAD_Bellenguez_noproxy_2026.*.gwas_finemap.rdsAD_Bellenguez_EADB_2022.*.gwas_finemap.rdsAD_Bellenguez_EADI_2022.*.gwas_finemap.rdsAD_Bellenguez_GRACE_2022.*.gwas_finemap.rdsAD_Bellenguez_main_2026.*.gwas_finemap.rdsAD_Bellenguez_nobiobank_2026.*.gwas_finemap.rds
- AD_GWAS_models › AD_TWAS_association › batch1 · syn75427328Joint analysis of eQTL + glycoQTL + pQTL; also listed on the other data type's pages
ROSMAP_Bennett_Klein_glycoQTL_Bellenguez.*.twas.tsv.gzROSMAP_eQTL_pQTL_AD_Bellenguez_2022.*.twas.tsv.gzROSMAP_eQTL_pQTL_AD_Bellenguez_EADB_2022.*.twas.tsv.gzROSMAP_eQTL_pQTL_AD_Bellenguez_EADI_2022.*.twas.tsv.gzROSMAP_eQTL_pQTL_AD_Bellenguez_GRACE_2022.*.twas.tsv.gz
- AD_xQTL_integration › AD_xQTL_colocalization_models › ROSMAP · syn74986543
ROSMAP.*.cb_xqtl_AD_Bellenguez.rdsROSMAP.*.cb_xqtl_AD_Bellenguez_EADB_2022.rds
- AD_xQTL_integration › AD_xQTL_colocalization_models › MSBB · syn74986553
MSBB.*.cb_xqtl_AD_Bellenguez.rdsMSBB.*.cb_xqtl_AD_Bellenguez_EADB_2022.rds
- AD_xQTL_integration › AD_xQTL_colocalization_models › KNIGHT · syn74986587
KNIGHT.*.cb_xqtl_AD_Bellenguez.rdsKNIGHT.*.cb_xqtl_AD_Bellenguez_EADB_2022.rds
- AD_xQTL_integration › AD_xQTL_colocalization_models › MiGA · syn74986600
MiGA.*.cb_xqtl_AD_Bellenguez.rdsMiGA.*.cb_xqtl_AD_Bellenguez_EADB_2022.rds
- AD_xQTL_colocalization_models › ROSMAP_glycoQTL › adjusted · syn76490248
adj_glyco_AD.*.cb_xqtl_AD_Bellenguez.rds
- AD_xQTL_colocalization_models › ROSMAP_glycoQTL › unadjusted · syn76490281
unadj_glyco_AD.*.cb_xqtl_AD_Bellenguez.rds
- ROSMAP_tpQTL › K5 › batch{1..2} · syn77716949
ROSMAP_tpQTL.*.cb_xqtl_AD_Bellenguez.rds
- AD_xQTL_integration › cTWAS_results › multi_group_cTWAS · syn74908390Joint analysis of eQTL + pQTL; also listed on the other data type's pages
ROSMAP_eQTL_pQTL.*.ctwas_multi_group_AD_Bellenguez_2022.tsv.gz
- AD_xQTL_integration › cTWAS_results › single_group_cTWAS · syn74908397
ROSMAP_eQTL_pQTL.*.AC_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.DLPFC_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.PCC_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.Ast_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.Exc_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.Inh_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.Mic_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.Oli_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.OPC_CUIMC1_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.monocyte_ROSMAP_eQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gzROSMAP_eQTL_pQTL.*.DLPFC_Bennett_pQTL.ctwas_single_group_AD_Bellenguez_2022.tsv.gz
Folders and files whose names carry this dataset's cohort, data type and tissue (bulk and single-nucleus data kept apart), taken from the staging folder list. All AD Knowledge Portal folders lists everything.
Contact
Oluwatosin Olayinka, Hao Sun and Rui Dong
Source files
The summary statistics are held on the Columbia University (CU) cluster. A related meta-analysis of GR@ACE, IGAP and UKB is publicly available in de Rojas 2021, Nature Communications.
Download source
The data are derived from the summary statistics of Bellenguez et al..
The cohort-specific data were obtained on request in early 2024 and uploaded to the Columbia University cluster in March 2024.
File Schema
chromosome: chromosome IDposition: hg38 positionref: hg38 reference allelealt: alternative effect allelevariant: variant ID in the formchr${chromosome}_${position}_${ref}_${alt}beta: SNP effect sizese: standard error ofbetapvalue: p-value ofbetaestimatemaf: minor allele frequencyn_cases: number of cases for estimaten_controls: number of controls for estimateoriginal_effect_allele_frequency: allele frequency of the original effect allele
Links to GWAS data analysis notebooks
- GWAS summary statistics processing: https://github.com/cumc/fungen-xqtl-analysis/blob/main/analysis/Wang_Columbia/GWAS/AD_GWAS_processing.ipynb lab access only
Cohorts included in this study
111,326 clinically diagnosed/proxy AD cases + 677,663 controls
- Stage I: 39,106 clinically diagnosed AD cases + 46,828 proxy-ADD cases + 401,577 controls
- Stage II: 25,392 AD cases + 276,086 controls
cohorts:
- EADB: The European Alzheimer & Dementia Biobank (15 European countries). Also called
EADB-TOPMedandEADB-core. 21,101,680 variants in sum.stats. - EADI: European Alzheimer’s Disease Initiative. 12,540,914 variants in sum.stats.
- UKBB: UK Biobank. We have the genotype and imputed variants ourselves.
- GR@ACE: the Genome Research at Fundació ACE. 61,744,410 variants in sum.stats.
- GERAD/PERADES
- DemGene
- Bonn
- the Rotterdam study
- the CCHS study
- NxC
- FinnGen:
- R6 includes 147,061 females and 113,344 males, and the one that Bellenguez paper used should be
R6/AD_WIDE(N=7,329 cases and N=252,879 controls). - R10 is the latest version (until 20240222), including 230,310 females and 181,871 males.
- R6 includes 147,061 females and 113,344 males, and the one that Bellenguez paper used should be
Supplementary Table 1: Demographic descriptions of the different meta-analyzed GWAS, including a description of EADB per country
| AD or proxy-ADD cases | Controls | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| N | % females | Age | Age at onset | APOE e4 allele frequency | N | % females | Age | APOE e4 allele frequency | ||
| DISCOVERY | ||||||||||
| EADB-TOPMed | 20,301 | 61.7 | 72.0±10.4 | 71.1±10.5 | 32.6 | 21,839 | 57.3 | 67.0±14.3 | 13.2 | |
| Belgium | 1,230 | 64.6 | 78.7±5.9 | 78.3±5.9 | 31.6 | 1,474 | 61.8 | 70.1±8.4 | 13.6 | |
| Bulgaria | 164 | 54.9 | 65.1±8.6 | 65.1±8.6 | 22.9 | - | - | - | - | |
| Switzerland | 182 | 64.3 | 76.0±6.5 | 76.9±6.0 | 19.2 | 388 | 55.9 | 74.8±4.0 | 10.1 | |
| Czech Republic | 183 | 60.7 | 75.8±7.8 | - | 31.7 | 61 | 65.6 | 66.9±7.2 | 10.7 | |
| Denmark | 403 | 57.1 | 79.6±7.8 | 79.6±7.8 | 33.7 | 654 | 54.4 | 73.1±8.5 | 15.4 | |
| Spain | 3,273 | 67.0 | 75.3±9.0 | 75.2±9.0 | 27.2 | 1,685 | 63.3 | 69.3±12.0 | 10.0 | |
| Finland | 1,151 | 64.0 | 70.9±8.8 | 69.8±8.5 | 42.0 | 1,806 | 51.4 | 71.8±7.1 | 15.9 | |
| France | 1,664 | 60.2 | 67.4±11.9 | 63.2±10.8 | 33.3 | 3,106 | 63.8 | 44.9±15.4 | 11.5 | |
| Germany | 1,628 | 60.3 | 74.8±9.4 | 74.6±9.8 | 33.1 | 2,050 | 56.0 | 74.2±8.0 | 12.3 | |
| Greece | 614 | 63.0 | 73.1±8.0 | 72.9±7.9 | 23.8 | 1,246 | 57.3 | 73.1±5.6 | 9.1 | |
| Italy | 3,271 | 68.1 | 73.7±8.9 | 72.2±8.7 | 25.0 | 1,317 | 56.8 | 72.2±10.5 | 8.6 | |
| The Netherlands | 2,438 | 55.8 | 66.2±10.7 | 65.6±10.5 | 41.9 | 2,389 | 47.5 | 60.1±12.0 | 17.9 | |
| Portugal | 80 | 75.0 | 69.9±9.2 | 69.2±8.9 | 30.0 | 74 | 75.7 | 67.2±6.8 | 17.6 | |
| Sweden | 1,533 | 62.9 | 72.8±11.2 | 72.8±11.2 | 40.7 | 3,089 | 61.8 | 70.6±9.8 | 15.6 | |
| United Kingdom | 2,487 | 51.1 | 68.0±10.7 | 66.4±10.1 | 34.4 | 2,500 | 51.8 | 74.4±7.2 | 12.8 | |
| EADB-HRC | 163 | 54.0 | 71.5±7.9 | 71.5±7.9 | 31.8 | 405 | 48.1 | 77.2±2.1 | 14.1 | |
| EADI | 2,400 | 65.6 | 74.3±10.1 | 73.9±10.2 | 29.4 | 6,338 | 60.3 | 80.0±7.6 | 10.5 | |
| GERAD | 3,030 | 63.2 | 78.1±9.3 | 78.1±9.3 | 35.1 | 7,153 | 52.0 | 50.7±11.7 | 15.4 | |
| Bonn | 635 | 65.5 | 77.8±9.8 | 77.8±9.8 | 30.1 | 1,210 | 54.8 | 69.9±9.3 | 12.6 | |
| RS1 | 1,165 | 72.9 | 83.7±0.2 | 83.7±0.2 | 33.4 | 4,739 | 56.7 | 82.8±0.1 | 12.9 | |
| RS2 | 141 | 59.6 | 82.8±0.6 | 82.8±0.6 | 27.1 | 1,961 | 54.1 | 73.3±0.2 | 14.1 | |
| GR@ACE/DEGESCO | 6,497 | 64.1 | 81.8±8.8 | 81.8±8.8 | 23.0 | 6,785 | 49.1 | 55.9±15.8 | 11.0 | |
| DemGene | 1,693 | 65.5 | 72.2±8.8 | 71.6±8.8 | 39.5 | 5,926 | 47.7 | 68.5±11.1 | 18.2 | |
| CCHS | 365 | 68.5 | 82.7±6.9 | 82.7±6.9 | 31.4 | 6,106 | 54.3 | 58.5±13.7 | 15.8 | |
| NxC | 269 | 72.4 | 78.7±6.9 | 78.7±6.9 | 26.0 | 675 | 44.4 | 51.9±8.9 | 10.0 | |
| UKBB-P | 49,275 | 57.3 | 59.3±6.7 | - | 22.6 | 338,440 | 56.0 | 55.8±8.2 | 14.0 | |
| REPLICATION | ||||||||||
| ADGC | 17,141 | 57.3 | 78.6±8.1 | 73.4±8.0 | 37.7 | 17,627 | 58.8 | 75.8±8.1 | 14.4 | |
| CHARGE | ||||||||||
| CHS | 450 | 66.0 | - | 81.9±5.2 | 34.0 | 1,702 | 60.0 | 81.1±5.0 | 20.0 | |
| FHS | 472 | 68.0 | - | 84.5±7.3 | 18.5 | 3,878 | 54.0 | 74.4±10.8 | 11.0 | |
| FinnGen | 7,329 | 48.8 | 82.2±7.4* | 78.0±7.6 | 31.6 | 252,879 | 56.7 | 59.3±17.4* | 17.8 | |
| *age at death or current age | ||||||||||
| total | 111,326 | 677,663 |
AD GWAS–xQTL Integration
AD GWAS–xQTL colocalization results (ColocBoost): syn69865816
AD GWAS–xQTL colocalization models: syn69670630
AD fine-mapping models (AD Knowledge Portal): syn69670625
Integrated AD Gene-Level Summary
AD risk genes prioritized by xQTL + GVC: syn70095142
AD risk genes prioritized by xQTL + TWAS + GVC: syn70095143
Bellenguez 2026 update (noproxy, main, nobiobank)
The 2026 files are fine-mapped from the summary statistics of the consensus meta-analysis of Alzheimer’s disease and related dementias (ADRD) by the EADB, PGC-ALZ and IGAP consortia: Bellenguez et al. 2026, Nature Genetics 58:1214–1225 (doi:10.1038/s41588-026-02583-1; preprint medRxiv 2025.10.20.25338060). The study reported 91 ADRD risk loci, 16 of them new. The fine-mapping for this release was run by Haochen and Yining.
Each analysis has a top-loci table (AD_Bellenguez_<variant>_2026.gwas.top_loci.bed.gz) and per-region SuSiE RSS fine-mapping objects (AD_Bellenguez_<variant>_2026.*.gwas_finemap.rds), staged on the AD Knowledge Portal alongside the 2022 release.
| Variant | Analysis in Bellenguez et al. 2026 |
|---|---|
main | Primary meta-analysis of European-ancestry samples from IGAP, EADB, PGC-ALZ, UK Biobank and FinnGen: 72,721 AD cases, 55,960 proxy ADRD cases, 614,267 controls and 235,566 proxy controls from 52 studies |
noproxy | The same meta-analysis without proxy ADRD cases and controls (people whose parent or sibling had dementia) |
nobiobank | The same meta-analysis without the large biobank samples, so mainly clinically diagnosed AD cases |
The paper does not list case and control numbers for the noproxy and nobiobank analyses in its main text; they are in its supplementary tables. The unified AD loci xQTL summary (unified_AD_loci_xQTL_summary_20261002.xlsx) integrates the 2026 loci with xQTL colocalization results.