All of Us finds many millions of new DNA variants as it moves toward 1 million participants from diverse backgrounds

by Jocelyn Kaiser

A giant research database with genetic and health data of people across the United States has hit its stride. Today, the All of Us study published an analysis of genomes and linked health information from nearly 250,000 participants whose diversity may be greater than that in any comparable biobank widely available for research. The National Institutes of Health (NIH)-funded effort, which plans to quadruple that genome number ultimately, has already found hundreds of millions of never-reported genetic variants that have the potential to reveal new links between genes, health, and disease.
“This is a hugely important paper because of the diversity that’s there and the resources it’s going to provide for researchers,” says human geneticist Rick Kittles of the Morehouse School of Medicine, who wasn’t involved in the analysis, published in Nature. The effort is already paying off: Two other papers also published today showcase how All of Us data can help develop genetic scores for predicting disease risks in diverse groups.
Launched in 2018, All of Us is NIH’s answer to similar projects in countries such as the United Kingdom, Iceland, and Japan. The gold standard is the 500,000-participant UK Biobank with its high-quality data available to any qualified researcher. But 94% of its participants are of European ancestry, limiting the relevance of findings to other groups.
All of Us aims to recruit 1 million people willing to share their DNA, blood samples, and electronic health records (EHRs). Diversity in many forms is a key goal: Of the 526,000 participants who have completed all enrollment steps, 47% are racial and ethnic minorities and about 80% are people over age 65, or who earn a low income, are disabled, or are from other groups that “historically have been left out of large-scale scientific studies,” says Alexander Bick, a geneticist with the All of Us data center at Vanderbilt University Medical Center.
That rich diversity in the project’s first 245,388 whole genomes revealed more than 1 billion genetic variants, more than 275 million of which were novel. This expands by roughly 30% the catalog of known variants found by the UK Biobank and other whole genome databases, says Bick, corresponding author on the Nature paper.
U.S. and international researchers, who can usually get approval within 1 day to access All of Us data using its cloud-based portal, have already begun to explore fresh links between disease and these variants. For example, a study last month used All of Us data to show that rare variants in a muscle gene that has been tied to heart disease in Europeans poses a similar risk to people of African ancestry. Identifying carriers of these genes in the clinic could help them take steps to reduce their risk.
Most of the 275 million new variants are rare, and many could prove harmless, which would allow clinicians to rule them out when searching for the gene behind a child’s disease, for instance. Other variants may cause functional changes in a gene’s protein that in turn trigger disease and could offer fresh targets for drug development.

Nearly half of the 526,000 people who are fully enrolled in the All of Us health and genomics study are nonwhite, and about 80% are disabled, older than age 65, sexual or gender minorities, or from other groups routinely left out of biomedical research.
D. AN-PHAM/SCIENCE
The other new studies that draw on All of Us data have used less-detailed genome scans of participants that look for common variants that slightly raise disease risks. Tallying hundreds or thousands of these markers to produce “polygenic risk scores” can identify people at high risk of say, heart disease or diabetes. One paper, in Nature, uses data from All of Us and other biobanks to firm up evidence that polygenic risk scores for type 2 diabetes cluster into groups depending on associated complications—some people have a signature linked to heart disease, others kidney disease for example, suggesting they may need different treatments for their diabetes. The other paper, in Nature Medicine, draws on All of Us data for 10 diseases to develop risk scores that work for people with African, Asian, and Hispanic ancestry. This “will help level the playing field” so not just white patients can learn their risks, says geneticist Alicia Martin of the Broad Institute.
Some of the more than 7000 researchers using All of Us are looking beyond genomics to explore questions about health care disparities. For example, the biobank’s detailed but anonymized survey data on the 10% of participants who report belonging to sexual and gender minorities is “an incredibly rich data source” that will enable studies of these often “unseen groups,” says social scientist Stephanie Cook of New York University. She is using it to explore whether homophobic discrimination and other stressors can raise LGBTQ individuals’ heart disease risk.
All of Us leaders note that the project also stands apart from other biobanks because participants can opt to receive insight from the program. About half of 51,000 people who were offered results back agreed to find out whether they carry risk genes for 59 diseases such as breast cancer and heart disease, Bick’s team reports. He says one future goal is to study “how returning this medical information to participants changes their health trajectories.”
The project has signed up 767,000 participants as of this week and hopes to reach 1 million by the end of 2026. “It’s still early days,” Martin says. “But "I’m really excited to see what comes from All of Us.”
doi: 10.1126/science.z6nhrfx