Science at this scale demands the highest standards.

Studies can originate in different ways:

1. A researcher at 23andMe initiates a study to answer an open scientific or health related question.

2. External researchers who are experts in specific fields can access 23andMe data to better understand the topic or condition they are studying.

3. An organization wants to recruit new participants to collaboratively advance research in a given area.

How a study begins

• 23andMe outlines the study plan.

• If collaborating, 23andMe enters into an agreement with the trusted 3rd party research organization.

• 23andMe draws on our existing pool of consented participants, or can recruit new participants who meet the study's criteria.

• Our team ensures that the study plan is incorporated in a research protocol.

• That protocol is reviewed by a third-party, Independent Ethics Review Board.

• Research is only conducted after such approval is received.

• The review includes not only the study plan but also the materials that will be sent to consented participants.

1. DNA, read once

   When a 23andMe member registers their 23andMe kit they submit their saliva sample and DNA is extracted. A customer can choose to consent to participate in research for the greater good. Their DNA is then de-identified and analyzed for research. We look at hundreds of thousands of specific positions in their genetic code. This happens once. The member does not need to do anything again for their genetic data to contribute to future studies.

   In some cases, new consenting participants are recruited for studies that focus on particular conditions. We may also collect additional types of samples, if needed for a specific study.

2. Health history, through surveys and more

   Research surveys ask about health conditions, medications, symptoms and lifestyle. The consenting member chooses which surveys to answer, and each one is clearly marked with the Research logo so they always know their responses may be used in a study. Participants may also choose to upload their medical records and other health information, including from devices and health trackers.

3. Genetic and phenotypic data combined

   Knowing someone's genetic code tells us about their biology. Knowing their health history tells us how that biology plays out in their life. Together, these two sources allow researchers to ask questions that neither could answer alone, which is what makes the 23andMe dataset unusually powerful.

1. Name and personal details are removed

   To protect member privacy, 23andMe conducts research with information that has first been stripped of identifying names and contact information. What remains is a set of de-identified genetic markers and health information.

2. Every data point is checked for accuracy

   Before an analysis begins, the genetic data goes through a series of quality checks — flagging samples that didn't read clearly, removing duplicates, and verifying the data is reliable enough to draw meaningful conclusions from.

3. We fill in the gaps — statistically

   Our array reads hundreds of thousands of genetic positions directly. Using a statistical technique, we can reliably infer millions more positions that weren't measured, by comparing one member's data to large reference datasets of known genetic patterns. This expands our ability to find meaningful associations without requiring anyone to be re-tested.

4. We look for patterns across hundreds of thousands of people

   The core of our research is comparing the DNA of people who have a particular condition — say, Parkinson's disease — to the DNA of people who don't. We're looking for genetic differences that appear consistently in one group. The larger the study, the more confidently we can distinguish a real signal from random noise. This is why scale matters so much, and why 11 million participants change what is scientifically possible.

5. We account for the fact that people are different in other ways too

   Age, sex and ancestry all influence health outcomes and can create misleading patterns in the data if not accounted for carefully. Our analyses include statistical controls for these factors so that what we find is genuinely about genetics, not about other characteristics our participants happen to share.

6. We run analyses separately across ancestry groups — by design

   Most genetic research has historically focused on people of European descent, which limits how broadly the findings apply. We run our analyses within multiple separate ancestry groups and combine the results — so our findings are more likely to hold up across diverse populations. This is one of the most important things that makes 23andMe research different and more impactful.

1. We don't publish the first thing we find

   Before we treat a finding as real, we confirm and validate it — running additional analyses, testing whether it holds up in different groups of people and checking whether independent datasets show the same pattern. Science that replicates is science worth trusting.

2. Then it goes through peer review

   Before publication, findings are reviewed by independent scientists in the field, just like at academic research institutions. These are experts who weren't involved in the research and have no stake in the outcome. This is the standard that all serious science is held to.

1. Findings are published and the data is made available

   When a study is complete, we publish the findings in a peer-reviewed scientific journal. We strive to make our publications open-access so people everywhere can learn and benefit from the results.

   We also make the underlying summary data publicly available, so that other scientists around the world can build on our work, run their own analyses and catch anything we might have missed. We believe that open science is better science, and that it's better for all of us as humans.

2. Participants receive new personalized insights based on the research

   When possible, if a finding is meaningful at the individual level — where knowing one's specific genetic result would genuinely be useful to them — we strive to return a personalized report back to the 23andMe research participant. These are clearly labeled as research-based, with guidance on what to discuss with a clinician. Consented participation creates a loop where value is often returned to the participant in the form of new insights they can use.