23andMe raw data: the complete guide to reading and analyzing it

The complete guide · Updated June 2026

23andMe & raw DNA

What is 23andMe raw data?

When 23andMe processes your sample, it doesn't sequence your entire genome. Instead it uses a genotyping chip to read roughly 600,000+ specific positions in your DNA called SNPs — single-nucleotide polymorphisms, the spots where one letter of the genetic code commonly varies between people. Your raw data is the complete table of every one of those markers: a plain text file you can download and analyze yourself.

The reports you see inside your 23andMe account interpret only a small, curated subset of those markers. The raw file contains all of them — which is why understanding 23andMe raw data lets you go much further than the standard reports. It is not a whole-genome sequence (that's a separate, far larger and more expensive test), but for looking up established variants, 600,000+ markers is plenty.

It helps to keep the scale in mind. Your genome is roughly three billion base pairs; a consumer genotyping array reads a few hundred thousand of them — a tiny fraction, deliberately chosen to capture common variation cheaply. That makes the file fundamentally a research-grade sample of your DNA, not a clinical read-out: it samples the genome where variation is most common and best understood, and stays silent everywhere else. Understanding what the file does not contain matters as much as what it does.

This is a hub guide: each section links out to a dedicated post. If you have an AncestryDNA or MyHeritage file instead, almost everything here applies — they export near-identical text files, and the same tools generally read all three.

How to download your 23andMe raw data

The download is done on the 23andMe website, not the mobile app. In short: log in on a computer, open Settings, find the 23andMe Data section, click Download Raw Data, verify your identity, and 23andMe will email you a link to a .txt file (sometimes inside a .zip).

For the full walkthrough with screenshots of each step, see our dedicated guide: How to download your 23andMe raw data.

The process is similar at other companies, though the menus differ. If your data lives elsewhere, follow the matching walkthrough: how to download your AncestryDNA raw data or how to download your MyHeritage DNA raw data. Either way you get the same kind of plain-text file, and everything below applies.

What's inside the 23andMe raw data file

Open the file in any text editor and you'll see a short header followed by one tab-separated row per marker. Each row has four columns:

• rsid — the reference ID for the marker (e.g. rs429358)
• chromosome — which chromosome the marker sits on (1–22, X, Y, or MT)
• position — the exact coordinate of that marker on the chromosome
• genotype — the two letters you inherited at that spot, one from each parent (e.g. AG)

A single line might read rs429358 19 44908684 TC. On its own that's unreadable — the meaning lives in what's known about each rsID. The rsID is a stable public reference number that lets researchers point to the same spot in the genome across studies and databases. The genotype is the pair of letters (A, C, G, or T) you carry at that position, one inherited from each parent; whether that pairing is common, rare, protective, or risk-associated is what the science has to tell you. To read 23andMe raw data in any useful way, you need a tool or reference that translates those genotypes into plain-language findings — and that translation is only ever as good as the research behind each marker.

How to interpret 23andMe raw data

There are three broad ways to interpret 23andMe raw data. You can look up individual markers by hand using public databases like SNPedia or ClinVar — accurate but slow. You can upload the file to a third-party web tool that generates a report — fast, but it means sending your genome to someone else's servers. Or you can analyze it on your own device, so the file never leaves your phone.

Each route has real trade-offs in cost, depth, and privacy. If you'd like to try the on-device route right now, you can read your 23andMe file in your browser with our free DNA explorer — nothing is uploaded. For a side-by-side comparison of the main options — free versus paid, upload versus on-device, and how deep each one goes — see the best tools to interpret 23andMe raw data.

Whatever tool you use, note the exact rsID behind any finding. That single ID lets you look the marker up independently in a public database and judge how strong the evidence really is — rather than trusting one tool's one-line summary.

What you can explore in your raw data

Broadly, the raw file opens up three kinds of exploration, each with its own confidence level.

Traits. These are the most fun and the lowest-stakes findings — things like caffeine metabolism, the cilantro-tastes-like-soap variant, muscle-fiber type, or whether you're likely to have wet or dry earwax. Even here, most traits are polygenic: shaped by many genes plus environment, so a single marker rarely tells the whole story. Lactose tolerance is a good illustration of how even a "simple" trait gets nuanced once you look closely — see is lactose intolerance genetic?.

Ancestry-linked context. Some variants are far more common in certain populations, and allele frequencies differ across ancestries. But it cuts both ways: many published associations were studied mostly in people of European descent, so a marker's real-world meaning can be weaker or simply unknown for other ancestries. Treat population context as one input, not a conclusion.

Carrier status and health-associated variants. This is where the raw file can look up markers tied to inherited conditions and disease risk — and where caution matters most. A carrier result or a flagged "risk" allele is an association drawn from population studies, not a diagnosis of you. Serious findings in this category need clinical-grade confirmation and, ideally, a conversation with a certified genetic counselor before you act on anything. For a grounded tour of which health signals are meaningful and which are noise, see what 23andMe raw data reveals about your health.

Specific markers people look up

A handful of variants drive most raw-data searches. Each deserves its own context, so we've covered them in depth:

• MTHFR — variants linked to folate metabolism, and why the results are often over-interpreted. See MTHFR in your 23andMe raw data.
• APOE4 — the marker associated with Alzheimer's risk, and why finding it in raw data calls for real caution. See APOE4 in your 23andMe raw data.
• BRCA — breast and ovarian cancer variants, and the serious limits of relying on a genotyping chip for them. See BRCA in your 23andMe raw data.
• Hemochromatosis — the HFE iron-overload variants and what carrier status actually means. See hemochromatosis in your 23andMe raw data.
• Blood type — how to estimate your ABO and Rh type from raw markers. See finding your blood type in 23andMe raw data.
• Factor V Leiden — the rs6025 clotting variant, and why it needs clinical confirmation. See Factor V Leiden in your raw data.
• COMT — the rs4680 "warrior vs worrier" dopamine variant. See COMT rs4680 in your raw data.
• CYP1A2 — rs762551, fast vs slow caffeine metabolism. See CYP1A2 caffeine metabolism.
• ALDH2 — rs671, the alcohol-flush variant. See ALDH2 in your raw data.
• ACTN3 — rs1815739, the "sprint gene" for power vs endurance. See ACTN3 in your raw data.
• CYP2C19 — how you metabolize drugs like clopidogrel and PPIs (one to confirm clinically). See CYP2C19 in your raw data.
• VDR — the vitamin D receptor variants (FokI, BsmI), and why they tell you less than the hype. See VDR in your raw data.

Accuracy and limitations of 23andMe raw data

This is the most important section to read. 23andMe raw data is generated for research and educational purposes — it is not clinically validated and is not a diagnostic test. Genotyping chips have known error rates, and a notable share of individual "risk" markers flagged in raw data turn out to be false positives when checked with a proper clinical method.

A few specific caveats:

• Coverage. The chip reads only the positions it was designed for, so it can miss disease-relevant variants entirely — a "normal" result is never an all-clear. A gene like BRCA has thousands of possible pathogenic variants and a consumer array checks only a handful; see BRCA in your 23andMe raw data for how badly this can mislead.
• Miscalls. Genotyping is a statistical measurement, and individual positions can be miscalled — reported as one genotype when the true one is another. A single line could simply be wrong, and no consumer tool can catch it.
• Polygenic reality. A listed genotype is an association, not a verdict. Most common conditions are polygenic and shaped heavily by environment and chance, so one SNP is a weak predictor at best. Pharmacogenomic markers like CYP2C19 show this: even where the genetics are clean, the clinical picture depends on which drug, which dose, and the rest of your health.

Treat anything you find as a starting point for conversation, not a conclusion. If you see a result that worries you — particularly around a serious condition like cancer risk or a degenerative disease such as the one associated with APOE4 — do not act on it. Confirm it with proper clinical-grade testing and discuss it with a doctor or a certified genetic counselor before drawing any conclusions.

Privacy and the safe way to analyze your raw data

Your genome is unusually sensitive information. It's effectively a permanent identifier — you can't change it the way you'd change a leaked password — and it doesn't only describe you. Because you share large stretches of DNA with your parents, siblings, and children, sharing your file partly exposes theirs, including relatives who never consented. That's what makes the decision of where your raw data goes weightier than for ordinary health data.

Most 23andMe interpretation tools ask you to upload your raw file to their servers. That means handing your genetic data to another company, subject to its policies, its security, and whatever happens to it if the company is breached or sold. Uploading your genome is close to a one-way decision, and it's worth pausing before you make it. The recent history of the industry has made this concrete — see what happens to your DNA data when a company is sold for why "we'll keep it safe" is a promise only as durable as the company making it.

There is a safer path: analyze the file on your own device, so it never leaves your phone. On-device parsing means the interpretation happens locally and nothing is transmitted — a fundamentally different privacy posture from cloud upload. For the full contrast between the two models, see on-device vs cloud health data privacy.

If you've already uploaded your data somewhere and want it gone, deletion is your other lever. It isn't always as simple as clicking a button, and it may not cover everything the company has already derived from your file, but it's worth doing — our guide walks through it: how to delete your 23andMe data.

How to explore your raw data responsibly

Pulling this together into a short playbook:

1. Download and keep your own copy. Store the file somewhere you control. Your account access can lapse or change; a local copy is yours.
2. Prefer on-device analysis. Where you have the choice, parse the file locally instead of uploading it. Free browser and app tools can do this without your genome ever leaving your device.
3. Start with low-stakes traits. Explore caffeine, cilantro, or muscle-type variants first to get a feel for how the file works before wading into health-associated markers.
4. Record the rsID behind every finding. It lets you verify any claim in a public database and judge the evidence for yourself.
5. Never self-diagnose. Remember the coverage gaps, the miscall risk, and the polygenic reality. A worrying result is a prompt to get clinical testing and talk to a genetic counselor — not a conclusion.

Done this way, your raw data becomes what it's actually good for: an interesting, private window into your own biology — one input among many, held to the right standard of proof and kept under your own control.

Frequently asked questions

What can my 23andMe raw data tell me?

It can let tools look up well-studied genetic markers across health risk, carrier status, medication response, and traits — far more than the standard 23andMe reports. It cannot diagnose anything, and it misses variants the chip wasn't designed to read, so treat results as educational, not definitive.

Is 23andMe raw data accurate?

It's accurate enough for ancestry and many traits, but it is not clinically validated. Genotyping chips have known error rates, and individual health-risk markers can be false positives. Any serious finding should be confirmed with clinical-grade testing and a healthcare professional.

Is it free to analyze 23andMe raw data?

Downloading your own raw data is free, and some tools interpret it at no cost while others charge. You can also analyze it on-device with an app like Quanome instead of paying a site to process and store your genome.

Is it safe to analyze my 23andMe raw data?

It depends on the method. Uploading your raw file to a third-party server exposes your genetic data to that company's policies and security. The safest approach is to analyze the file on your own device, where it never leaves your phone.

Does the raw file cover my whole genome?

No. A genotyping chip reads a few hundred thousand hand-picked positions out of roughly three billion base pairs — a tiny fraction of your genome. It samples common variation well but skips most rare and disease-relevant variants entirely, which is why a 'normal' raw-data result is never an all-clear.

Can I use AncestryDNA or MyHeritage raw data the same way?

Yes. AncestryDNA and MyHeritage export very similar tab-separated files with the same core columns — rsID, chromosome, position, and genotype — so the same interpretation tools generally work. The exact markers each chip reads differ slightly, so a variant present in one file may be missing from another.

Should I trust a health-risk marker I find in my raw data?

Treat it as a lead, not a diagnosis. A listed genotype is a statistical association, not a verdict, and consumer chips can miscall individual positions. For anything serious — cancer-risk or neurodegenerative variants especially — confirm with clinical-grade testing and talk to a certified genetic counselor before drawing conclusions.

Why is my DNA data considered so sensitive?

Your genome is effectively a permanent identifier you can never change, and because you share large stretches of it with your relatives, sharing your data partly exposes theirs too. That's why where you send the file — and whether you can delete it later — matters more than for ordinary health data.