The report said SNP, and it sounded like news
You were not looking for a new vocabulary. You opened a file, a relative forwarded a screenshot, or a company highlighted a row in red. The label was SNP. It looked medical. It felt personal.
A SNP, pronounced snip, is a single nucleotide polymorphism. In plain language it is a place in the genome where people disagree by one DNA letter. One person has an A. Another has a G. The rest of the nearby sequence is the same.
Most SNPs are not accidents that happened to you last year. They are old spelling differences, copied faithfully, sitting in millions of other genomes. The interesting question is rarely whether a SNP exists. It is what, if anything, that one letter does.
One letter in a three-billion-letter text
DNA is a four-letter code: A, T, C, and G. Those letters pair in a fixed way, A with T and C with G, which is how the double helix holds together. Human nuclear DNA is about three billion of those letters long. About 20,000 of the stretches are protein-coding genes.
A SNP is a substitution at one of those positions. MedlinePlus describes it as a difference in a single DNA building block. If two people lined their genomes up, most letters would match. About one in a thousand would not.
That density is why a person typically carries millions of SNPs relative to the reference sequence. A one-letter change can sit inside a gene, near a gene, or in the long stretches of DNA that do not code for protein. Location is not a verdict, but it is the first fact worth knowing.
Why SNPs are common, not rare events
Polymorphism is the giveaway in the name. A SNP is common enough to show up again and again. Older definitions asked for a frequency of at least 1 percent. Rare one-letter changes exist too. They are often called single nucleotide variants, and they can matter in disease. They are not what a spit-kit chip was built to harvest.
Common SNPs persist because most of them do little harm. A letter can change without rewriting the protein or breaking a switch. A few change a protein in a way that history has already tested, for better, worse, or both depending on the environment.
That is why SNPs became the workhorse of modern genetics. If a spelling is common, you can count it in thousands of people and ask whether it shows up more often with a trait. The SNP is often a marker for a neighborhood of DNA, not the cause itself.
- SNP: one-letter site that is common in a population
- SNV: any one-letter change, including rare ones
- Indel: extra or missing letters, which is a different kind of variant
- Most SNPs: no known effect on health
rsIDs are public names, not the letters you carry
Once a SNP is seen often enough to catalog, databases give it a public name. In dbSNP that name is an rsID, a Reference SNP cluster ID. rs334, rs1800562, and rs429358 are names of sites. They are not names of people.
The rsID tells you which street corner you are standing on. It does not tell you which way you walk. Your genotype is the pair of letters you carry at that corner, one from each parent for nuclear DNA. The public record describes the site. Only a local file, a clinical assay, or a carefully called sequence describes you.
ClinVar may attach a submitted clinical label to the same rsID. gnomAD may say how often each letter appears in sequenced people. Those are facts about the variant in the world. They are not a sentence about your future.
A genotyping chip samples hundreds of thousands of SNPs, not three billion bases
This is the gap that consumer reports quietly step over. A 23andMe or AncestryDNA export is a list of predetermined sites, typically on the order of 600,000 rows. The chip does not walk the genome letter by letter. It asks, at each designed probe, which of two common spellings is present.
Three billion bases is the book. A few hundred thousand SNPs is a set of bookmarks, chosen because they are common, cheap to assay, and useful for ancestry matching. They were not chosen to complete a medical record.
A site that is not on the chip is not wild type. It was not called. Whole-genome sequencing tries to read nearly all of the letters. An exome tries to read the protein-coding 1 percent. A consumer chip is neither. If a report implies that your spit kit scanned every gene, it is describing a test you did not take.
- Consumer chip: hundreds of thousands of common SNPs
- Exome: protein-coding sequence, about 1 percent of the genome
- Whole genome: an attempt to read nearly all three billion bases
- Missing chip row: not called, not normal
Three SNPs people actually meet
rs334 sits in HBB, the gene for beta-globin, on chromosome 11. One letter change can swap a glutamic acid for a valine, the classic sickle cell allele. Homozygous sickle cell disease and heterozygous sickle trait are different clinical objects. Looking up rs334 is a clean way to see how SNPs and disease can meet. It is not a hemoglobinopathy assay.
rs1800562 sits in HFE. Older papers call it C282Y. It is discussed in hereditary hemochromatosis, a recessive iron-overload condition, and it is not rare in people of European ancestry. A heterozygous carrier is not the same as a person with iron overload. Ferritin measures iron. An rsID does not.
rs429358 sits in APOE. Together with rs7412 it helps define the common haplotypes labeled ε2, ε3, and ε4. ε4 is a well studied risk factor for late-onset Alzheimer disease at the population level. Risk is not a forecast. Querying only rs429358 is incomplete for haplotype calling. APOE status is sensitive. Keep the letters on the machine that already holds them.
Inside a gene, between genes, and what risk is allowed to mean
Most SNPs sit between genes. They can still be useful as landmarks. A genome-wide association study finds SNPs that are slightly more common in people with a trait, then points researchers toward a region. The SNP in the headline is often a signpost, not the mechanism.
When a SNP falls inside a gene it can be silent, change one amino acid, or rarely wreck a splice site. Missense SNPs are easy to print. Easy to print is not easy to interpret. A changed amino acid can be tolerated in millions of people, or it can matter in one biochemical context and not another.
Risk language is a population sentence. It says that, on average, people with this spelling have a different chance of a trait. It does not say what will happen to the person reading the row. A SNP can be a clue. It is not a crystal ball.
What a SNP cannot tell you
A SNP cannot tell you that you have a disease. Disease is a clinical conclusion. A SNP cannot tell you that a missing chip probe is the healthy spelling. Missing is a gap. A SNP cannot tell you the rest of a gene that the chip never touched.
It cannot tell you zygosity unless a file or an assay called the site. It cannot turn a submitted ClinVar word into a care plan. It cannot make a consumer kit into whole-genome sequencing by putting the letters in a nicer PDF.
The honest use of a SNP is narrower and still valuable. You can learn the public name. You can learn whether the site is common. You can learn whether labs have submitted an interpretation, and how contested that interpretation is. Then you can stop before the story becomes about you in a way the data does not support.
Look the public name up. Leave the file where it is.
The practical next step is not a new upload. The rsID is already on the internet. dbSNP places the site. ClinVar may hold a submitted interpretation. gnomAD may hold a frequency. Those pages answer questions about the variant, not about a person.
If you have a raw-data file, the private fact is the two letters next to the rsID, or the fact that the row is missing. That file does not need to travel for you to read the public record. Identifier lookup and genotype matching are different jobs.
This page is not medical advice. Clinical decisions need a clinician and an assay that can actually see the variants that matter. Once you know that a SNP is a named one-letter site, the next literacy is the name itself.
Questions
What is a SNP?
A single nucleotide polymorphism: a one-letter difference in DNA at a site where both spellings are seen in people. It is a variant site, not a diagnosis.
What does SNP stand for?
Single nucleotide polymorphism. Pronounced snip. Nucleotide means one DNA letter, A, T, C, or G. Polymorphism means the difference is common enough to recur in a population.
Is a SNP a mutation?
It is a one-letter change, and older language called many such changes mutations. NIH pages now prefer variant, because most SNPs do not cause disease. Mutation vs variant is a wording choice with a lot of baggage.
How many SNPs does a person have?
MedlinePlus estimates roughly 4 to 5 million SNPs in a person's genome, about one every 1,000 letters. Consumer chips measure only hundreds of thousands of predetermined sites, not all of those differences.
What is an rsID for a SNP?
An rsID is dbSNP's public name for a variant site, such as rs334 or rs429358. The name identifies the corner. Your genotype, if a file called it, is the pair of letters you carry there.
Do DNA tests read all of my SNPs?
No. A consumer chip samples a designed list of common SNPs. It is not whole-genome sequencing and not an exome. Sites that were never on the array are not called.
What are rs334, rs1800562, and rs429358?
Public examples. rs334 is the well known HBB sickle site. rs1800562 is HFE C282Y. rs429358 tags common APOE haplotypes with rs7412. Looking them up is not a diagnosis.
Can a SNP diagnose a disease?
No. A SNP is a spelling at a site. Clinical diagnosis needs a clinician, the right assay, and the rest of the evidence. This explainer is not medical advice.