How to read your raw DNA file — rsid, chromosome, position and genotype
Open a raw DNA file expecting something impenetrable and you find a text table. Roughly 650,000 rows, four columns, no encryption and no special format. Anyone can open it in Notepad or TextEdit.
Understanding those four columns is enough to look up any marker in your own genome and read what the research says about it. This is how.
What do the columns in a raw DNA file mean?
Four columns, in this order: an rsid naming the marker, the chromosome it sits on, its position along that chromosome, and your genotype — the two letters you inherited there. Some services split the genotype into two columns. Everything else in the file is header comment lines.
A few rows from a 23andMe export look like this:
# rsid chromosome position genotype
rs4477212 1 82154 AA
rs3094315 1 752566 AG
rs4988235 2 136608646 GG
rs762551 15 75041917 AC| Column | What it is |
|---|---|
rsid | The catalogue number for this position, shared across every service |
chromosome | 1–22, plus X, Y and MT for mitochondrial DNA |
position | Where on that chromosome the marker sits, counted in bases |
genotype | The two letters read at that position — your result |
The lines beginning # at the top are comments. They record which reference genome build the positions refer to, which matters: a position number is only meaningful relative to a build, and builds 37 and 38 use different coordinates for the same marker.
What is an rsid?
A reference SNP identifier — the catalogue number for one position in the human genome, assigned by dbSNP at the US National Center for Biotechnology Information. It looks like rs4988235. Every service uses the same rsids, which is what makes files from different companies comparable at all.
This is the single most useful thing in the file. The rsid is the key that connects your data to the published literature: paste one into dbSNP, ClinVar or the GWAS Catalog and you get every study that has looked at that position.
It also means a marker keeps its name forever. rs4988235 is the same position whether it appears in a 23andMe file from 2012, an AncestryDNA file from last week, or a paper published in 2004.
Why does each marker have two letters?
Because you have two copies of every chromosome — one from each parent — so every position carries two readings. AG means one copy reads A and the other G. AA or GG means both copies match. The file does not say which parent contributed which letter.
Two consequences worth knowing.
First, markers on the X chromosome in males, on the Y chromosome, and in mitochondrial DNA appear as a single letter, because there is only one copy to read. You will see rows like rs2534636 Y 2655180 T.
Second, some rows read -- or 00. That is not a result — it means the array failed to call that position on your sample. A few thousand no-calls in a file of hundreds of thousands is completely normal.
How do you look up one of your own markers?
Open the file in a text editor, search for the rsid, and read the genotype in the last column. Then look the same rsid up in dbSNP or ClinVar to see what published research associates with it. That is the whole mechanism — no special software required.
If your file arrived as a .zip, unzip it first. If it is a .csv from MyHeritage or FamilyTreeDNA, the separators are commas and the values are usually in quotes, but the columns mean the same things.
One caution about searching: check the strand. Some services report the genotype relative to the forward strand of the reference genome and some relative to the marker’s own orientation, so a marker published as C/T may appear in your file as G/A. Comparing your file directly to a paper without checking this is the most common way an amateur lookup goes wrong.
What can you not learn from reading the file yourself?
Context. A single genotype means little without knowing how common it is, which populations the research covered, how strong the association was, and whether other markers modify it. Reading one line is easy; weighing it against the literature is the part that takes work.
A genotype that appears in 40% of people is not a finding about you. An association reported once in a study of two hundred people, never replicated, is not a finding either. And an association established in one population may have no evidence behind it in another — most published genetics research was done in European cohorts, which is a real limitation rather than a caveat added for form.
That gap between reading a line and knowing what it supports is the whole of the work. The file is transparent. The literature around it is not.
What we look at in a raw DNA file →