rsIDs, i-numbers, and How Variants Get Named

Scroll through your raw DNA file and one column is full of codes like rs4988235. They are not random. Each is a name assigned to a specific variant, and knowing how those names work turns your file from a wall of letters into something you can actually look up.

What an rsID is

An rsID - short for “reference SNP cluster ID” - is a stable public identifier for a variant, assigned in a database called dbSNP. When researchers submit a newly observed variant, dbSNP gives it a number prefixed with rs, such as rs4988235. From then on, that identifier points to the same variant everywhere: in your raw file, in scientific papers, in reference databases. It is the shared name that lets different tools and studies talk about the exact same position without ambiguity.

Because the rsID travels with the variant rather than with any single file, it stays constant even when position numbers change between genome builds. That stability is exactly what makes it useful as a lookup key.

How a variant earns its name

A variant gets an rsID once it has been observed and submitted to dbSNP. The database clusters together submissions that describe the same position and orientation, then assigns a single rs number to represent them. This is why the “cluster” is in the name - one rsID can gather many independent reports of the same variant into one canonical entry.

The number itself carries no meaning. A low number is not more important than a high one; it usually just means the variant was catalogued earlier. Think of rsIDs like catalogue numbers in a vast library rather than a ranking.

Why some rows have i-numbers instead

Not every position in a consumer file carries an rsID. Companies sometimes use their own internal identifiers for variants. A common example is the “i” numbers used by 23andMe, which look like i5000980 rather than rs.... These internal identifiers show up in two main situations:

• The variant does not yet have an rsID in dbSNP, so the company assigns its own label.
• The company is using its own custom probe or an internally tracked position that it prefers to name itself.

An i-number is not inferior data - it is simply named in a private scheme rather than the public one. The trade-off is that you cannot always look it up in a public database the way you can an rsID, because the mapping lives with the company.

Names can change

Public identifiers are not frozen forever. As dbSNP is updated, rsIDs can occasionally be merged or retired. If two separate rsIDs turn out to describe the very same variant, one may be merged into the other, and the retired number is redirected to the surviving one. Sometimes an entry is withdrawn entirely if it turns out to have been an error.

For everyday exploration this rarely causes trouble - databases keep records of merges so an old rsID still leads you to the current entry. But it is worth knowing that a name in an older file might have been superseded, which is one reason the same variant can appear under slightly different labels across tools.

Using the name to look things up

The practical payoff of understanding naming is that you can investigate a single row yourself. Once you know the identifier in your file, you can search a reference database to learn what is documented about that variant - which gene it sits near, how common each allele is, and what associations have been studied.

If you want to find a specific variant in your own export first, a plain text search does the job:

grep '^rs4988235' my_raw_dna.txt

That returns the line if the variant is present, showing your genotype alongside its chromosome and position. To understand the other columns sitting next to that name, see our tour of a raw file line by line, and for a gentler primer on the variants themselves, read what SNPs are.

Knowing how variants are named is one of the small skills that makes your raw file feel legible rather than cryptic. You can explore all of those named positions privately with on-device DNA analysis that keeps your file in your browser.

This article is educational and is not medical advice.