A consumer DNA test doesn’t read your whole genome. It checks a chosen set of spots, a few hundred thousand to a few million, where human DNA commonly differs by a single letter. That makes it well suited to questions about where your family came from and who you’re related to. Health is another matter: what the test “reveals” is often a statistical nudge or even a mistake, not a diagnosis, and findings can be false.1PubMed Central. Direct-to-Consumer Genetic Testing Who you are matters too, because results are most reliable for people of European ancestry, the group most research has studied.2PubMed Central. Analysis of polygenic risk score usage and performance in diverse human populations
What an SNP Actually Is
DNA is spelled with four chemical letters: A, C, G and T. An SNP, pronounced “snip,” is a spot where one letter differs between people, for example a C in some people and a T in others.3MedlinePlus. What are single nucleotide polymorphisms (SNPs)? SNPs are the most common kind of genetic difference between people.4MedlinePlus. What are single nucleotide polymorphisms (SNPs)? A typical person differs from the standard reference genome at 4 to 5 million places, nearly all of them SNPs or tiny insertions and deletions.5Wikipedia. Single-nucleotide polymorphism
Your genome is about three billion letters long, so those differences are a small fraction of it. Most sit outside genes and change nothing you can see. You inherit one copy of each spot from each parent, so you can carry two different letters at the same place. That pair of letters is what a test reports as your result.
A result of C/T means you carry a C on one copy of that spot and a T on the other. T/T means both copies match. The letters mean nothing on their own until researchers have linked that spot to something, so most entries in your file have no interpretation attached.
Most SNPs are ordinary human variation, not defects. Whether one matters depends on where it sits and what it does.
How the Test Reads Your DNA
Consumer tests use a genotyping chip, which checks pre-selected spots rather than reading every letter.6PubMed Central. Direct-to-Consumer Genetic Testing The chip is dotted with tiny DNA probes, each matching one version of a particular SNP.7GeNotes. Microarray At the lab, your DNA is copied to make enough of it, broken into fragments, and tagged with glowing dye.8PubMed. SNP array The fragments stick to the probes that match your letters, and a scanner measures the glow at each spot.9GeNotes. Microarray Software then turns the pattern of glow into a list of which letters you have at each spot.10GeNotes. Microarray
At each spot, the scanner measures how strongly the dye glows for each of the two possible letters, so the result is a ratio rather than a yes or no. A strong glow for one letter means two copies of it, while roughly equal glow for both means one of each. When the glow is muddy and doesn’t fall clearly into either pattern, the spot is usually reported as a “no call” and left blank instead of guessed. That’s why a file can contain a few blank entries even when the test worked. If a sample gives too weak a signal, the lab often reruns it or asks for a new kit. You end up with interpreted reports and, usually, a downloadable file of your letters.
Chips typically check about 500,000 to 1 million spots.11PubMed Central. How SNP Genetic Testing Works and What It Reveals Software can estimate many unmeasured SNPs by comparing your pattern with fully sequenced reference genomes, because nearby letters tend to be inherited together.12PubMed Central. How SNP Genetic Testing Works and What It Reveals So some results you see were predicted, not directly measured.
The chip also only finds what it was built to look for. A spot not printed on it can’t appear in your results, and a rare harmful change in your family may be missing from the chip entirely.
Ancestry, Relatives and Traits
Patterns of SNPs across the genome record the history of human migration and mixing.13Egyptian Journal of Medical Human Genetics. Single nucleotide polymorphism in genome-wide association of human population: A tool for broad spectrum service Even a handful of well-chosen SNPs, as few as 10, can tell apart continental origins.14PubMed Central. Proportioning whole-genome single-nucleotide-polymorphism diversity for the identification of geographic population structure and genetic ancestry Methods can also estimate which ancestry each stretch of your DNA came from, which is useful for people with mixed backgrounds.15PubMed Central. Estimating local ancestry in admixed populations
Neighboring regions, such as two nearby countries, share most of their history, and their SNP frequencies overlap heavily. That is why finer regional claims come with wide margins of uncertainty, and why broad continental results stay steadier than country-level percentages. Your percentages can also move between companies or after an update, because results compare you to reference groups. Different groups give different numbers, and adding more reference samples can shift your breakdown.
You inherit DNA in long chunks from each parent, and each chunk can be traced to a population it resembles. Chunks are reshuffled every generation, so the amount of DNA from a specific distant ancestor shrinks with each step back. A distant ancestor can leave no detectable DNA in you at all, even though they are your ancestor on paper. That reshuffling is also why two full siblings can get noticeably different ancestry percentages.
Shared SNP stretches can reveal relatives as distant as several generations back.16PubMed Central. Law enforcement use of genetic genealogy databases in criminal investigations: Nomenclature, definition and scope Close matches are rarely wrong, while distant cousin estimates are rougher. Matching can also uncover unknown siblings, donor parents or adoptions, so consider whether you want to see matches before turning the feature on.
Among traits, eye color is strongly linked to a few SNPs in and near the OCA2 and HERC2 genes, which separate blue from brown well.17PubMed Central. Forensic DNA Phenotyping: Genes and Genetic Variants for Eye Color Prediction In-between colors like green and hazel are harder to predict.
What It Reveals About Health Risk
Large research studies use common SNPs to find genome regions linked to common diseases like heart disease or diabetes.18Single Nucleotide Polymorphisms. GWAS to Identify SNPs Associated with Common Diseases and Individual Risk: Genome Wide Association Studies (GWAS) to Identify SNPs Associated with Common Diseases and Individual Risk A polygenic risk score adds up the small effects of many such SNPs into one number.19PubMed Central. Power and Predictive Accuracy of Polygenic Risk Scores But in one breast and prostate cancer analysis, scores sorted people barely better than a coin flip, around 53 to 56 percent.20PubMed Central. Power and Predictive Accuracy of Polygenic Risk Scores That fits what researchers think many risk SNPs do: they probably turn genes up or down rather than breaking them, so each has a small effect.21PubMed Central. Making sense of GWAS: using epigenomics and genome engineering to understand the functional relevance of SNPs in non-coding regions of the human genome
When you see a “slightly increased risk” result, ask what it is compared with, which is usually the average risk in your group. A risk score counts only the common SNPs researchers have studied, so it leaves out rare variants, family history, age and the environment. Someone with a low score can still develop the disease, because common conditions have many causes. A reassuring result is no reason to skip recommended screenings.
A single well-studied variant, such as a known harmful change in a specific gene, can raise risk far more than a risk score can. Even then, not everyone with such a variant develops the disease, and a chip can detect only the specific changes it was built to check.
Ancestry changes how informative scores are. Most research behind risk scores studied people of European ancestry.22PubMed Central. Selection, optimization and validation of ten chronic disease polygenic risk scores for clinical implementation in diverse US populations Scores are about one-third as informative for people of African ancestry as for Europeans.23PubMed Central. Analysis of polygenic risk score usage and performance in diverse human populations
Carrier results work differently. A carrier has one copy of a variant for a condition like cystic fibrosis, which could affect children if the partner is also a carrier. Yet a common 23-variant cystic fibrosis panel missed 31% of couples at risk in one large screening program.24PubMed Central. Sequencing as a first-line methodology for cystic fibrosis carrier screening Such panels caught about 96% of carriers of Northern European ancestry but only 64% of South Asian ancestry.25PubMed. Cystic fibrosis carrier screening in Australia: comparing sequencing and targeted panels across diverse ancestries
Some SNPs in liver enzyme genes change how fast you process certain medicines, affecting dose and side effects.26PubMed Central. Polymorphic Cytochrome P450 Enzymes (CYPs) and Their Role in Personalized Therapy Some of these variants are common in one population and nearly absent in others.27PubMed Central. The genetic landscape of major drug metabolizing cytochrome P450 genes—an updated analysis of population-scale sequencing data Many drug-response genes also have rare variants a chip won’t include. Never change a dose on your own based on a consumer result; share it with your doctor or pharmacist, who can order a clinical test.
Why Health Findings Often Turn Out Wrong
Most companies let you download your raw data, a file listing your letters at every spot tested.28PubMed Central. Third-Party Genetic Interpretation Tools: A Mixed-Methods Study of Consumer Motivation and Behavior Many people then upload it to free third-party sites for health reports. But 40% of variants in consumer raw data were false positives when rechecked by a clinical lab.29Genetics in Medicine. False-positive results released by direct-to-consumer genetic tests highlight the importance of clinical confirmation testing for appropriate patient care In another study, only 3 of 12 raw data file results suggesting a harmful variant were confirmed.30PubMed Central. Direct-to-Consumer Genetic Testing Most of those errors came from third-party sites interpreting the file.31PubMed Central. Direct-to-Consumer Genetic Testing Some variants such sites labeled “increased risk” were classed harmless by clinical labs and are common in the population.32Genetics in Medicine. False-positive results released by direct-to-consumer genetic tests highlight the importance of clinical confirmation testing for appropriate patient care
Several things feed the problem. Chips are tuned for common variants seen in many people, and rare variants give weaker, less reliable signals on them. Clinical tests confirm results with a second, more exact method. Third-party sites take your file and match each spot against public lists of reported variants. Many entries on those lists were labeled harmful from small or early studies and later downgraded. The sites usually don’t re-check whether the chip read that spot accurately, so an error in the file passes straight into the report, often in alarming language with no explanation of the uncertainty. People sometimes undergo unnecessary medical tests to chase these results.33PubMed Central. Direct-to-consumer raw genetic data and third-party interpretation services: more burden than bargain?
A variant that many healthy people carry is unlikely to cause a rare severe disease, because it would then show up in many more patients. Clinical labs weigh how common a variant is when deciding if it is harmful. If a site flags a frequent variant as a risk, it is often reflecting a weak statistical link rather than a diagnosis.
If a scary variant shows up, treat it as unconfirmed until a clinical lab retests a fresh sample. Ask your doctor for a referral to a genetic counselor, and tell the counselor which site produced the result and bring the file. Avoid major decisions like surgery based on raw data alone.
SNP Chips Versus Whole Genome Sequencing
Whole genome sequencing reads roughly three billion DNA letters, versus a chip’s hundreds of thousands to millions of spots.34SelfDecode. Confused About Whole Genome vs SNP Array Testing? Here’s What You Actually Need It reads the letters in order across nearly the entire genome instead of checking only spots chosen in advance. It can therefore find a rare change nobody has listed before, including one in your own family, and detect missing or repeated pieces of DNA that a chip may only hint at.
Sequencing-based carrier screening found at-risk couples a targeted panel missed.35PubMed Central. Sequencing as a first-line methodology for cystic fibrosis carrier screening In medical testing of chromosome changes, sequencing matched chip results and pinpointed them more precisely with fewer false alarms.36PubMed. A Comparative Study of Medium-Coverage Genome Sequencing and SNP Array Technology in Identifying Chromosomal Abnormalities to Advance Prenatal and Postnatal Diagnosis
Cost is also shifting. Very light sequencing now costs about $30 per sample in research labs, below a typical chip at about $39.37PubMed Central. Ultra Low-Coverage Whole-Genome Sequencing as an Alternative to Genotyping Arrays in Genome-Wide Association Studies But “very light” means each stretch of DNA is read only a few times, giving a thin view of the genome. Software fills in the gaps using reference genomes, much as chip results are filled in, so a low price doesn’t mean the full detail of a clinical sequencing test.
For ancestry and relatives, a chip test gives what most people want. If you have a family history of inherited disease, ask a doctor about clinical sequencing. Consumer sequencing still needs clinical confirmation before medical use, and more data also means more uncertain findings to interpret.
Who Else Can Use Your DNA Data
Police have uploaded crime-scene DNA to public genealogy sites like GEDmatch and FamilyTreeDNA.38PubMed. Using genetic genealogy databases in missing persons cases and to develop suspect leads in violent crimes Matches are usually cousins, so they build family trees to work out which relative fits the age and location of the suspect. Using this approach, they identified suspects and missing people in over fifty cold cases in one year.39PubMed. Using genetic genealogy databases in missing persons cases and to develop suspect leads in violent crimes The suspect doesn’t need to have tested, because a relative’s file is enough to narrow the field. This raises ethical and legal questions that remain unsettled.40Forensic Genomics. Investigative Genetic Genealogy: Ethical, Legal, and Social Issues and Directions for Future Research
Rules for consumer testing are still taking shape. Proposals include proof that tests are clinically valid, marketing limits, and doctor involvement.41PubMed Central. Considerations for developing regulations for direct-to-consumer genetic testing: a scoping review using the 3-I framework Third-party upload sites raise added privacy concerns.42PubMed Central. Third-Party Genetic Interpretation Tools: A Mixed-Methods Study of Consumer Motivation and Behavior
Policies differ between companies. Some databases let users decide whether to be included in law-enforcement matching, while others don’t allow it at all. Policies can change, and a company can be sold or go bankrupt, which can change who controls the data. Read the privacy terms before you test, not only after, and opt out of law-enforcement matching if it’s offered.
You can usually request deletion of your data and sample. Remember that relatives share your DNA, so your choice affects them too.