· 5 min read · Reviewed by the DNA Info Lab editorial team
23andMe Raw Data File Explained: Format, Columns & Limits
What is the 23andMe raw data file? A tab-separated text file of about 650,000 rows with four columns — rsid, chromosome, position and genotype. It records ~0.02% of your genome at carefully chosen positions. Below: what each field means, the genome build, no-calls, and the limits.
What the file actually is
When you download your data from 23andMe you get a zipped .txt file. Unzipped, it is plain text: a short header of comment lines starting with #, then one row per genetic position the chip measured. A typical file has around 650,000 rows and is only a few megabytes — because it stores letters, not the whole genome.
The four columns
| rsid | chromosome | position | genotype |
|---|---|---|---|
| rs4988235 | 2 | 136608646 | AG |
| rs6025 | 1 | 169519049 | TT |
| rs429358 | 19 | 45411941 | CT |
- rsid — the variant's reference ID in dbSNP (e.g. rs429358). Some rows use an internal
i-prefixed ID for 23andMe-specific probes. - chromosome — 1–22, plus X, Y and MT (mitochondrial).
- position — the base-pair coordinate on that chromosome, in the file's genome build (see below).
- genotype — the two letters (alleles) you carry at that position, one from each parent, e.g.
AG.
Genome build: GRCh37 / hg19
The position numbers are relative to a reference build. 23andMe files use GRCh37 (hg19). This matters if you cross-reference positions with a database on a different build (e.g. GRCh38) — the coordinates will not line up. Matching by rsid avoids the problem, which is why most tools (including ours) key on rsid.
No-calls: -- and 0
Not every position reads cleanly. When the chip cannot call a genotype it writes -- (23andMe) or 0 (AncestryDNA). A small fraction of no-calls is normal and simply means that position is skipped in any analysis.
Why it is not your whole genome
A microarray genotypes specific, pre-chosen positions — about 650,000 of the genome's ~3 billion base pairs (~0.02%). Those positions are picked to cover the most-studied, disease-associated common variants, so the file is very useful for risk and trait analysis, but it will not contain rare private mutations that were not on the chip. For that you need whole-genome sequencing.
How to open it
It is just text — any editor opens it, though 650,000 rows can be slow in a word processor. To actually interpret it, either look up individual rsids by hand in SNPedia/dbSNP, or upload the whole file to a tool that cross-references it against curated research.
Frequently asked questions
Can I open the 23andMe raw file in Excel?▼
Yes — it is tab-separated text. But with ~650,000 rows it may be slow, and Excel can mangle some values. A plain text editor or an analysis tool is safer.
What genome build does 23andMe use?▼
GRCh37 (also called hg19). Match variants by rsid rather than position to avoid build mismatches.
What do -- or 0 mean in the genotype column?▼
They are no-calls: the chip could not reliably read that position. They are skipped in analysis and a small number is normal.
Is the raw file my complete DNA?▼
No. It covers ~0.02% of your genome — the pre-selected positions on the chip. It is enough for common-variant risk and trait analysis but not for rare private mutations.
See what your file contains
Upload your 23andMe, AncestryDNA or MyHeritage raw file and get a free report that reads every relevant position for you.
Related articles
This article is for educational purposes only and does not constitute medical advice, diagnosis or treatment. Always consult a healthcare professional for decisions about your health.