Landmark 137 · Research infrastructure
23andMe and UK Biobank cohorts
In short
Consumer genetics (23andMe) and a 500,000-person biobank gave Parkinson's genetics many times more people than clinics could recruit, and let researchers study risk before diagnosis.
- What it is
- Research infrastructure
- Outcome
- Confirmed and built on
- When
- 2019–2020
- Research lane
- Infrastructure & funding
- Last checked
- 20 Oct 2026
The research record
This is the project's file for this landmark, as written and checked. Sources are listed at the end; anything not yet confirmed is marked "Not yet verified".
Who
The 23andMe Research Team and the International Parkinson's Disease Genomics Consortium (Mike Nalls and colleagues); UK Biobank researchers including Jacobs, Belete, Noyce and colleagues.
What happened
- 23andMe and UK Biobank in the GWAS. The 2019 meta-analysis of landmark 19 used 37,688 cases, 18,618 UK Biobank "proxy" cases (people with a parent who has Parkinson's) and about 1.4 million controls, finding 90 risk signals. The 2019 paper used 23andMe's earlier Parkinson's web-study data (Chang 2017) plus newer participants, and 23andMe scientists are co-authors; the share of the counts is not given in the abstract.
- UK Biobank on its own. Jacobs et al. (JNNP 2020;91:1046–1054, DOI 10.1136/jnnp-2020-323646): 1,276 incident Parkinson's cases among 500,406 controls. Risk rose with family history of Parkinson's or dementia, non-smoking, low alcohol, depression, daytime sleepiness, epilepsy and earlier menarche. People in the top 10% of a genetic risk score had odds 3.37 times those in the bottom 10% (95% CI 2.41–4.70). The risk score also improved a prediction model only modestly.
- Access. The 2019 paper's data statement says 23andMe summary statistics are available only to qualified researchers under an agreement with the company that protects participant privacy.
Why it moved us toward a cure
Sheer size finds weak genes and tests risk factors. It also provides a pool for finding people at high risk before symptoms, who are the likely participants for prevention trials.
Outcome
Advanced. Proxy cases and self-reported diagnoses are less precise than clinic-confirmed ones.
Sources
- Nalls et al. 2019 record (UK Biobank publication list)
- Jacobs et al., JNNP 2020;91:1046–1054, DOI 10.1136/jnnp-2020-323646 (PMID 32934108; one commentary, no correction found)
- Nalls et al. 2019 full text (data access statement), PMC8422160
- Not yet verified: the exact 23andMe share of cases and controls; 23andMe's own publications; funders of the UK Biobank analysis.
How it connects
Built on
- builds on 90 genetic risk signals · 2019
Led to
None recorded.