Can You Find Your Blood Type in Raw DNA Data?

How the ABO blood group is written into three variants in your raw file, what DD at rs8176719 means, and why Rh status is the part SNP chips cannot see.

“What is my blood type?” is one of the most common questions people bring to a raw DNA file, and it is one of the few that a consumer file can actually answer - with a caveat or two. ABO type is set by a single gene, the differences between A, B and O come down to a handful of variants, and most genotyping chips read the important ones. Rh status is another story.

Three letters, one gene

The ABO gene on chromosome 9 makes an enzyme that decorates the surface of red blood cells with a sugar. The A version of the enzyme attaches one sugar, the B version attaches a slightly different one, and the O version is broken and attaches nothing. Your type is simply which working versions you carry: two O copies makes you type O, an A with anything makes A (unless a B is also present, which makes AB), and so on.

Yamamoto and colleagues worked out the molecular basis in 1990, and it is remarkably tidy:

  • O is caused by a single-base deletion near the start of the gene (c.261delG). It shifts the reading frame and the enzyme never gets made.
  • B differs from A by a few amino-acid changes, two of which do most of the work: c.796C>A and c.803G>C.

Those three changes are all single-position variants, which means they have rsIDs, which means a SNP chip can read them.

The variant that tells you whether you are type O

The O deletion is rs8176719. Because it is an insertion/deletion rather than a letter swap, raw files report it with the indel convention: D for the deleted (O) allele and I for the intact allele.

  • DD - two O alleles, so you are type O.
  • DI - one O allele and one working allele. You are A or B, and the other variants decide which.
  • II - two working alleles: A, B or AB.

23andMe files have included this marker for years. It is worth checking that it appears in your file before trusting the result, because not every chip version reads it, and a marker that is absent is not the same as a marker that says “II”. If it is missing, the rest of this method cannot tell O from A or B.

Telling A from B

Once you know how many working alleles you have, rs8176746 and rs8176747 separate A from B. On the forward strand - the way raw files report - the less common allele at each marks the B version: T at rs8176746 and G at rs8176747. The two travel together, so you would normally see them agree.

Putting it together for someone with DI at rs8176719:

  • If both A/B markers show only the common allele, the working copy is A, so the type is A.
  • If both show the less common allele, the working copy is B, so the type is B.

And for II:

  • Both markers common only: AA genotype, type A.
  • Both markers heterozygous: one A and one B, type AB.
  • Both markers showing only the less common allele: BB, type B.

There are rarer ABO alleles (A2, various weak subgroups) that this shortcut ignores, so a small share of people will be mislabelled. For most, the three-variant reading matches their blood-bank card.

Why Rh status does not work the same way

The “positive” or “negative” after your letter is the Rh D antigen, and it is set by a different gene, RHD. Most Rh-negative people of European ancestry are negative because the entire RHD gene is deleted. A SNP chip cannot read a gene that is not there; it only sees that its probes for that region returned nothing, which looks the same as a failed probe. Some services estimate Rh status from nearby variants that tend to travel with the deletion, and the estimates are reasonable in some populations and poor in others. Treat any Rh guess from raw data as exactly that.

Checking it yourself

You can search your file for the three rsIDs with a text search, or load it into an on-device analysis that reads the file in your browser. If you use the command line:

grep -E '^rs8176719|^rs8176746|^rs8176747' my_raw_dna.txt

Two cautions. First, some services do not report indels at all, so rs8176719 may simply be absent. Second, do not use a result derived this way for anything where blood type actually matters - transfusion, pregnancy planning, or a medical form. Those situations use a serological test on actual blood, which takes minutes and is definitive. The raw-file version is a satisfying piece of self-knowledge, not a clinical record.

If you want to go deeper on how variants get their identifiers and why a file sometimes lists an “i” number instead of an rs number, our guide to how variants are named covers it.

References

This article is educational only and is not medical advice. Blood type for any clinical purpose must come from a laboratory blood test.

Further reading