Which 23andMe Chip Version Is My Raw Data From and Does It Matter
Your 23andMe raw file name carries the chip version. Look for v3, v4 or v5 in the .txt or .zip name, then check whether a missing SNP was a probe gap or a no-call.
Your 23andMe raw data file name carries the chip version. Open the .txt or .zip name and look for v3, v4 or v5 after your name, like genome_John_Doe_v5_Full_20211128000000.txt. The v3 chip reads about 967,000 markers, v4 about 570,000 and v5 about 640,000. For most people the version matters less than it seems. A missing SNP is usually a chip design choice, not a failed sample.
Where the v3, v4 or v5 is written
Look at the file name before you open the data. 23andMe names the download with your name, the chip version, and a date stamp.
- Download the raw data zip from 23andMe.
- The zip name looks like genome_YourName_v5_Full_20211128000000.txt.zip.
- Unzip it. The .txt inside has the same name.
- The token v3, v4 or v5 is the chip version.
If someone renamed the file, open the .txt and count the data lines. v3 has about 967,000, v4 about 570,000, v5 about 640,000. That is a slower way but it works when the name has been changed. Do not worry if your file has slightly fewer lines. 23andMe removes probes that did not pass quality control. The Full in the name means all genotypes that passed quality control, not a whole-genome sequence.
Which 23andMe chip version is my raw data from and does it matter
It matters in a narrow way. The version controls which markers were on the array, not whether the file is valid.
The chip version changes how many markers are present. A missing rsID is usually a probe that was never on that array.
The practical effects:
- v3 is the largest raw file, with roughly 967,000 markers. It is older and has fewer of the newer health-specific probes.
- v4 is the smallest common raw file, at about 570,000 markers.
- v5 sits in the middle at about 640,000 markers but added targeted probes for some health and drug-response variants.
Larger is not automatically better. A v5 file can answer a question about a specific health variant that an older v3 file may not cover, even though v3 has more total lines.
Why a missing SNP on one chip is usually not an error
A genotyping chip has fixed probes. If the probe for an rsID is not printed on the v4 chip, no line for that rsID can appear in the file. Your DNA can be perfectly fine. The marker was just never measured.
That is different from a marker that was tested but failed quality control. In the raw .txt file, search for the rsID.
| What you see | What it means | |
|---|---|---|
| Marker not on chip | no line at all | Chip did not test that marker |
| Marker tested but no-call | line with -- in genotype | Signal below quality threshold |
Scroll the table sideways for the rest →
In a 23andMe .txt file, search for an rsID.
When a third-party report lists a SNP as missing, check the raw file. If the line is absent, your chip did not measure it. If the line is there with -- in the genotype column, that one call failed. The common mistake is treating both as one problem. They are not the same. If a tool reports "not genotyped" rather than "normal", that is usually the chip version showing up.
What the chip difference actually changes
The changes are specific.
What changes: - Which rsIDs are present. - How many markers a tool can read directly. - Whether a rare variant or a custom probe is covered.
What does not change: - The validity of your file. - The quality of the markers that are present. - The ability to upload a .txt or .zip to most DNA tools.
Most common trait and ancestry reports use markers shared across v3, v4 and v5. Missing one obscure marker rarely changes a result. In practice, a v4 file is smaller but still enough for most consumer reports. A v5 file may fill one health-related gap. A v3 file may have extra ancestry markers. None of them is broken.
Does a v3 file still work?
Yes. A v3 file from a decade ago still parses in tools that accept 23andMe raw data. If a particular finding needs a marker that version lacks, that one line is skipped. The file itself is not rejected.
The data columns are the same: rsID, chromosome, position and genotype. That is why old files remain readable. If you are comparing your file with a friend's v5 file, expect different marker lists. The overlap is large but not perfect. Some tools also impute missing common markers, meaning they infer them from a reference panel of sequenced genomes. That is less certain than direct genotyping but useful for polygenic scores.
What to do next
The download steps for 23andMe, on the page itself.
The short version: what is worth acting on, first.
To see this in your own numbers, upload the same 23andMe file at free DNA analysis. It reads v3, v4 and v5 raw files in the browser and reports traits, medication response and carrier-relevant findings. The trait story appears before you give an email, and ancestry is optional.
New to raw DNA files? Start with the full guide: How to Download & Analyze Your 23andMe Raw Data Free
Written by
HealthOS Research