What SNPs Are and Why They Matter for DNA Testing

Open any consumer DNA test’s technical FAQ and the same three-letter acronym keeps appearing: SNP. AncestryDNA says it genotypes more than 700,000 SNPs. 23andMe lists hundreds of SNP-based health reports. MyHeritage matches you to relatives based on shared SNP segments. If SNPs are doing this much work, it is worth understanding what they actually are.

The short version: SNPs are the few hundred million positions in the human genome where people commonly differ from each other by a single letter. Consumer DNA tests read a curated subset of them.

A SNP, defined without jargon

DNA is a long sequence of four chemical letters, A, C, G, and T, repeated about three billion times across your 23 pairs of chromosomes. Most of those positions are identical between any two humans. At a small fraction of positions, roughly 1 in every 300 to 1,000 letters, people in the population reliably differ.

A position where one common version of the letter (say, an A) and another common version (say, a G) both appear in the population, each in a meaningful fraction of people, is called a Single Nucleotide Polymorphism. “Single nucleotide” means one letter. “Polymorphism” means multiple forms.

Each SNP is identified by a stable ID number like rs53576 or rs429358. Geneticists catalog these in public databases such as dbSNP, maintained by the US National Center for Biotechnology Information.

Why ~700,000 SNPs is enough

Consumer ancestry tests genotype somewhere between 600,000 and 900,000 SNPs, typically called “700K” as shorthand. That is a tiny fraction of the three billion letters in your genome but it is plenty for the questions these tests answer.

Two reasons:

• Linkage disequilibrium. SNPs near each other on a chromosome tend to be inherited together in large blocks. Reading one SNP in a block tells you a lot about the rest. A well-designed 700K chip captures most of the common variation across the genome by tagging these blocks rather than reading every position.
• Common variation is enough for population comparisons. Ancestry inference, relative matching, and most consumer health reports work on common SNPs, the ones that appear frequently enough in different populations to be statistically informative. You do not need to read rare variants to estimate someone’s continental ancestry.

This is why AncestryDNA, 23andMe, and MyHeritage can deliver a usable ancestry breakdown from a sub-$100 chip rather than a multi-thousand-dollar whole-genome sequence.

How SNPs power ancestry

Different ancestral populations developed slightly different SNP frequencies over thousands of years. A particular SNP might appear in one form 80 percent of the time in northern European populations and only 20 percent of the time in East Asian populations.

When the company’s pipeline compares your roughly 700,000 SNP calls against reference panels of people with known ancestry, it is matching your pattern of common-versus-rare letters at each position to the patterns typical of each population. That is what produces the percentages on your ethnicity dashboard.

The accuracy of any ancestry estimate depends on how representative the reference panels are. Brands with broader and more diverse reference panels generally produce more confident calls in underrepresented regions.

How SNPs power relative matching

You inherit half your DNA from each parent in long segments. Two relatives share unbroken stretches of identical SNP calls along their chromosomes wherever they inherited the same DNA from a common ancestor. The longer and more numerous the shared segments, the closer the relationship.

Matching pipelines scan your 700K SNP profile against every other profile in the database, flag segments longer than a threshold (typically around 7 centimorgans for a confident match), and estimate the relationship from the total amount shared. This is how AncestryDNA can place a stranger as your third cousin from a saliva tube.

What SNPs do not tell you

SNP genotyping reads pre-selected common positions. It does not read rare or novel variants, structural changes like large deletions, or repetitive regions like the ones causing Huntington’s disease. For those, you need whole-genome or whole-exome sequencing. We cover the differences in Genotyping vs Sequencing Explained.

This also matters for health reporting. Consumer health reports built on SNP chips can flag a fixed set of well-studied variants. They cannot screen comprehensively for every disease-relevant change. The US Food and Drug Administration has authorized specific direct-to-consumer reports on this basis. Any consumer report is a starting point for a conversation with a clinician, not a clinical-grade diagnostic.

If you are weighing what a flagged SNP result might mean for you, the National Society of Genetic Counselors directory is the right next step.

Where this fits

SNPs are the basic unit consumer DNA tests are built on. Knowing they exist as a curated subset of common variation explains both what these tests do well and where their limits sit. For more on what each test reports, see our Guide to DNA Testing. For the lab process that produces these SNP calls, see How DNA Test Labs Process Samples.