Learn Guide
What to do with your raw DNA data, and what the file cannot settle
The raw DNA data file from 23andMe, AncestryDNA or MyHeritage is the same file whatever you do with it next. It can be kept, read again for medication and trait results, used to place your maternal and paternal lines, and uploaded to a genealogy database to find more relatives. It cannot settle a rare disease-causing variant, and that limit matters more than any of the uses.
Key takeaways
- Download the file and keep your own copy. The route at eight companies is in our download guide, and the file works as it arrives, zipped or not.
- A chip file can answer some medication questions. The ten exports we measured let our report read between 1 and 12 of its 14 medication positions[1].
- A rare disease-causing variant flagged in raw data is more likely wrong than right: sequencing confirmed 16 percent of chip calls at variants rarer than 1 in 100,000, and 4.2 percent for disease-causing BRCA1 and BRCA2 variants[2].
- GEDmatch, Living DNA and FamilyTreeDNA each say on their own pages that they take in files from other companies and that the upload is free[3, 4, 5].
- Carrier status and the full clinical panel need a sequenced genome. No chip file can give them reliably.
Raw data is the uninterpreted file behind a DNA test's reports: one line for each position the company's chip measured, with the two letters it found there. The ten chip exports we have measured held between 563,320 and 955,958 lines[1]. Chips read common variants accurately and rare ones poorly. In 49,908 UK Biobank participants with both chip data and exome sequencing, sequencing confirmed 99.0 percent of chip calls at variants more common than 1 in 100, and 16 percent at variants rarer than 1 in 100,000[2].
Keep your own copy
The first thing to do with raw data is to have it. The file sits in your account at the company that tested you, and it stays reachable only as long as the account and the company do. Our download guide has the route at eight companies, each read on the company's own help page. Some links expire within a day and some files take a week to prepare, so it is worth doing before you need the file.
Keep the archive as it arrived. There is no need to unzip, rename or convert it: every use below takes the file as the company delivered it. Once it is on your computer the company's security no longer covers it, so store it where you would store any other health record.
Read it again for medication and trait results
The company that tested you reported on part of what its chip measured. The rest of the positions are still in the file, and the databases they can be read against have moved since your test. A re-analysis reads the same file against current releases.
How much a chip file can answer depends on which positions that chip carried. We ran ten real chip exports through our report's own reader. Of the 1,572 single positions the report scores, they read between 20 and 37 percent, and of the 14 positions behind the medication cards they read between 1 and 12[1].
20 to 37%
of the positions our report scores, read by ten real chip exports[1]
12 of 14
medication positions read by the best exports[1]
1 of 14
read by the weakest[1]
Our chip report can read up to seven medication genes in full, and gives a result for a gene only when the file read every position that result depends on. The medication report guide lists the genes and shows, export by export, which files read them. Traits and polygenic scores come from common variants, which is what chips are built for; a score is given as a band, and polygenic scores in plain words explains what the band means.
Check a rare health flag before you believe it
Many raw data tools will list every position in your file that some database links to a disease. For rare variants those lists are mostly wrong, and the reason is the chip. Weedon and colleagues compared chip calls with exome sequencing in 49,908 UK Biobank participants. For variants rarer than 1 in 100,000, sequencing confirmed 16 percent of the chip's positive calls. For disease-causing BRCA1 and BRCA2 variants it confirmed 4.2 percent[2].
The opposite reading is also unsafe. A raw data file with no flagged variant has not cleared anything, because a chip measures only the positions it was built with. This is why our chip report leaves carrier status and rare clinical findings out.
Place your maternal and paternal lines
Two parts of the genome pass down one line of the family each: mitochondrial DNA through the mother's mother's line, and the Y chromosome from father to son. A file that carries enough of their marker positions can be placed on a branch of each tree, called a haplogroup. What a haplogroup is covers what the name does and does not say about ancestry.
Whether a chip export carries those positions depends on the company and the year, and a file from someone without a Y chromosome has no paternal line to place. The file check says which lines your own file can place.
Upload it to a genealogy database to find more relatives
Relative matching works only inside one company's database, so people tested elsewhere never appear in your match list. Some databases take in files from other companies for that reason. The table gives three, each described from its own page as we read it on 17 November 2026. Their rules change, and the company's page is the one to trust.
| Database | Files it says it accepts | What it says about cost |
|---|---|---|
| GEDmatch | DNA data from "all major DNA testing companies"; its upload page names 23andMe, FTDNA and AncestryDNA[3] | A free account; it describes itself as "a free, public DNA database"[3] |
| Living DNA | AncestryDNA, 23andMe, MyHeritage, FamilyTreeDNA (Family Finder), Gene by Gene and Genographic (Geno 2.0)[4] | The upload is free, with a paid upgrade for full autosomal results[4] |
| FamilyTreeDNA | AncestryDNA, 23andMe and MyHeritage, for the chip versions its page lists[5] | The upload and DNA matches are free, with a paid upgrade for further tools[5] |
Source: Each company's own page, read 17 November 2026[3, 4, 5]. A service is listed only if its page could be read that day.
Matching is a different use from a report, and it has a different privacy cost. A report is read by you. A matching database exists to show your relatives that you are in it, and to show you to them. Read what each one says about who can see your profile, and about law-enforcement access, before uploading.
Know what the file cannot do
A chip file cannot give carrier status or a clinical-grade answer about a rare variant, for the reason in the section above[2]. It cannot count copies of a gene, so it cannot answer CYP2D6; the medication report guide shows why. And where the chip did not measure a position, no software can read it.
Those questions need a sequenced genome. A whole-genome VCF or gVCF covers the positions a chip skipped, and a gVCF also records where the sequencing looked and found the reference, so a report can say a gene was examined. VCF, gVCF, BAM and CRAM explains the files a sequencing company hands over.
Before you upload the file anywhere
A raw data file identifies you and, in part, your relatives. These questions are worth asking of any service that takes it, ours included.
- Where is the file processed and stored, and under which country's law?
- Is the file deleted after the analysis, or kept? If kept, for what?
- Is the data sold, shared with researchers, or used to train anything, and can you say no?
- Does each result name the source and version it came from?
- What does the service do with a position your file lacks? One that fills the gap with the common version will print a normal result for a gene it never saw.
- Can you see what your file can answer before you pay?
For our own report: the file is processed and stored in Finland and deleted after analysis by default, every result names its source, and the file check reads the file inside your browser, uploads nothing, and shows what a report from it would cover. A chip report costs €19 in Europe and $19 elsewhere, paid once.
What Aimosti would (and wouldn't) show you
From a chip export our report reads Factor V Leiden and prothrombin G20210A, up to seven medication genes, traits, ancestry, the maternal and paternal lines where the file carries them, and polygenic scores given as a band. Each result names its source, and a position the file did not read is listed as not read. The file check shows which of these your file can fill before you pay.
What we won't claim
We won't report carrier status or a rare clinical finding from a chip file, read a missing position as a normal result, or describe another company's upload rules from anything but its own page. We do not match relatives. Nothing in a raw data report is a diagnosis, and a result that matters should be confirmed by a clinical test.
Bottom line. A raw data file is worth keeping and worth reading again. It answers common-variant questions well: several medication genes, traits, ancestry, the two single-line haplogroups. It answers rare-variant questions badly, so a rare health flag from any raw data tool is a reason to ask for a clinical test and nothing more.
Questions people ask
What can I do with my raw DNA data?
Keep a copy, read it again for medication genes, traits and ancestry, place your maternal and paternal lines, and upload it to a genealogy database to find more relatives. It is a poor tool for rare disease-causing variants: a chip flag for one of those is more often wrong than right[2].
What can I do with 23andMe raw data that 23andMe did not report?
The file holds every position the chip measured, including ones the company's reports did not use. A re-analysis reads them against current databases. In the three 23andMe exports we measured, our report could read 11 or 12 of its 14 medication positions[1].
Is it safe to upload raw DNA data to a website?
It depends on the site. Ask where the file is processed and stored, whether it is deleted afterwards, and whether the data is sold or shared. A check that runs in your own browser, as ours does, uploads nothing at all.
Can raw DNA data tell me about my health?
In part. A chip file can answer common-variant questions such as several medication genes and polygenic scores. It cannot be trusted with rare variants: sequencing confirmed 16 percent of chip calls at variants rarer than 1 in 100,000[2]. Carrier status needs a sequenced genome.
Which sites accept raw DNA data uploads for finding relatives?
GEDmatch, Living DNA and FamilyTreeDNA each say on their own pages that they accept files from other testing companies and that the upload is free, as read on 17 November 2026[3, 4, 5]. Check each company's page for the file versions it takes.
Do I need a new DNA test to get more from my data?
Not for what a chip can answer: the file you already have can be read again. For carrier status, rare variants or CYP2D6 you need a sequenced genome, which is a different test.
References
- What your DNA file can actually read: measured on 14 real files. Aimosti, 2026. Measured 6 October 2026; lists the ten public chip exports by vendor and year.
- Weedon MN, Jackson L, Harrison JW, et al. Use of SNP chips to detect rare pathogenic variants: retrospective, population based diagnostic evaluation. BMJ, 2021. doi:10.1136/bmj.n214 Tables 1 and 2 and the abstract: 49,908 participants; calls confirmed 99.0% above a frequency of 1 in 100 and 16.1% below 1 in 100,000 (Axiom chip); 4.2% for pathogenic BRCA1 and BRCA2 variants.
- GEDmatch: home page and How it works. GEDmatch, 2026. Read 17 November 2026. Home page: compatible with DNA data from all major DNA testing companies; a free, public DNA database. How it works: generic uploads (23andMe, FTDNA, AncestryDNA, most others).
- Free DNA upload. Living DNA, 2026. Read 17 November 2026. Accepts DNA data from AncestryDNA, 23andMe, MyHeritage, FamilyTreeDNA (Family Finder), Gene by Gene and Genographic (Geno 2.0); the upload is free, with an option to pay for an upgrade to full autosomal results.
- Autosomal DNA upload. FamilyTreeDNA, 2026. Read 17 November 2026. Accepts raw DNA data from AncestryDNA, 23andMe and MyHeritage for the array versions listed on the page; no cost to upload and see DNA matches, with an optional paid upgrade.
Last reviewed . Every number on this page links to the source it comes from; if one of them has moved, tell us.
Free · in your browser
Find out what your own file can read.
The free check opens your DNA file on your device and shows which format you have and what a report could read from it. The file is never uploaded.