DNA analysis results can look overwhelming at first, full of letter codes, percentages, and risk scores that seem to require a genetics degree to parse. The good news: once you understand a few core concepts, you can read your results with real clarity. Whether you’re looking at ancestry estimates, health risk reports, or raw data files, the key is knowing what the numbers actually mean and where their limits are.
What Your Raw Data File Contains
If you download your raw data from a company like 23andMe or AncestryDNA, you’ll get a text file with hundreds of thousands of rows. Each row represents a single spot on your genome where people commonly differ from one another. These spots are called SNPs (single nucleotide polymorphisms), which are just locations where one “letter” of your DNA code varies between people.
Each SNP has three pieces of information you need to understand:
- RSID (Reference SNP ID): A unique label starting with “rs” followed by a number, like rs1234567. This is the universal name for that specific spot on your genome, used across all databases and research studies. If you want to look up what a particular SNP means, the RSID is what you search for.
- Chromosome and position: The physical location of the SNP on one of your 23 chromosome pairs.
- Genotype: Two letters (like AA, AG, or GG) representing the two copies of DNA you inherited, one from each parent. Each letter is called an allele. If both letters are the same (AA), you’re homozygous at that position. If they differ (AG), you’re heterozygous.
Most of these SNPs have no meaningful health or trait implications on their own. The ones that do are the basis for the interpreted reports your testing company generates.
How to Read Ancestry Percentages
Ancestry results are estimates, not measurements. Your report might say you’re 54% English or 12% West African, but those numbers come from statistical models that compare your DNA to reference populations. They are not as precise as they appear.
Every ancestry percentage has a confidence range behind it. AncestryDNA, for example, might report 54% England and Northwestern Europe, but the actual confidence interval could be 52% to 69%. That means the company’s best guess is 54%, but the true proportion could reasonably fall anywhere in that range. On 23andMe, the default confidence level for reporting an ancestry category is just 50%, which the company itself labels “speculative.” At that threshold, a prediction is as likely to be accurate as it is to be wrong.
If your results include the option to adjust confidence levels, try sliding it up to 80% or 90%. You’ll notice some categories shrink or disappear entirely, and a larger chunk of your ancestry gets labeled “Broadly European” or “Unassigned.” That’s the honest version of your results. The more specific a geographic prediction tries to be (country level versus continent level), the less reliable it becomes, because neighboring populations share enormous amounts of DNA.
Three Types of DNA Tests
Not all DNA tests look at the same parts of your genome, and the type of test determines what your results can tell you.
Autosomal DNA tests are the standard offering from 23andMe and AncestryDNA. They examine DNA from all 22 non-sex chromosome pairs plus the X chromosome, capturing contributions from both parents. These are best for ethnicity estimates, finding relatives, and identifying health-related SNPs. Their reach for ancestry is roughly five to seven generations back before the signal gets too diluted.
Y-DNA tests trace the strict paternal line: father to grandfather to great-grandfather and so on. They assign you to a paternal haplogroup, a branch on the human family tree defined by shared mutations. Only biological males can take a Y-DNA test, since the Y chromosome passes exclusively from father to son.
Mitochondrial DNA (mtDNA) tests trace the strict maternal line. Mitochondrial DNA passes from mother to child regardless of sex, so anyone can take this test. Like Y-DNA, the result is a haplogroup assignment showing your deep maternal ancestry.
If your report mentions a haplogroup, it’s describing ancient migration patterns tens of thousands of years ago. It tells you where your direct paternal or maternal line originated, not your overall ethnic makeup.
Interpreting Health Risk Reports
Health reports from consumer DNA companies typically fall into two categories, and reading them correctly requires understanding the difference.
Single-Gene Variants
Some reports test for specific, well-studied mutations with strong effects. The FDA has authorized 23andMe to report on selected variants in genes related to breast and ovarian cancer (three BRCA1/BRCA2 variants most common in people of Ashkenazi Jewish descent), late-onset Alzheimer’s disease, Parkinson’s disease, celiac disease, blood clotting disorders, and a handful of other conditions. These are the most straightforward results to read: you either carry the variant or you don’t.
But “straightforward” doesn’t mean “complete.” The BRCA report, for instance, tests only three of the thousands of known BRCA mutations. A negative result does not mean you’re free of BRCA risk. It means you don’t carry those three specific variants. The same limitation applies to every FDA-authorized health report on these platforms.
Polygenic Risk Scores
For common conditions like heart disease, type 2 diabetes, or depression, your risk isn’t driven by a single gene. It’s influenced by hundreds or thousands of small-effect variants spread across your genome. Companies combine these into a polygenic risk score that places you on a bell curve relative to other people.
These scores show relative risk only. A score in the top 10% for type 2 diabetes means your genetic predisposition is higher than 90% of the comparison group, but it doesn’t tell you your actual likelihood of developing the disease. That absolute risk depends on factors the DNA test can’t capture: your weight, diet, activity level, family history, and age. A woman carrying a BRCA1 mutation has a 60% to 80% absolute risk of breast cancer. A high polygenic risk score for a different condition might translate to an absolute risk increase of just a fraction of a percent.
There’s another important caveat. Most genomic studies have been conducted on people of European ancestry, so polygenic risk scores are most accurate for that population. If your background is East Asian, African, South Asian, or Indigenous, the scores may be less reliable or essentially uninformative.
Relative Risk vs. Absolute Risk
This distinction is the single most important concept for reading any health-related DNA result. A report might say a variant increases your risk of a condition by 50%. That sounds alarming until you learn the baseline. If the average person’s risk is 1 in a million, a 50% increase brings it to 1.5 in a million. The relative risk jumped by half, but the absolute risk barely moved.
Whenever you see a percentage increase or decrease, ask: increase relative to what? If your report doesn’t state the baseline population risk, look it up. A 2x relative risk on a condition that affects 0.01% of people is very different from a 2x relative risk on something that affects 10% of people. The first is still extremely unlikely. The second is genuinely worth paying attention to.
Using Third-Party Analysis Tools
Many people upload their raw data to independent tools for deeper analysis. Promethease is the most widely used health-focused option, with 97% of its users reporting they received information about specific disease risks. GEDmatch is the go-to for extended genealogy and finding genetic relatives across different testing platforms. Other tools like Genetic Genie and Livewello focus on specific health pathways.
If you use Promethease, your results will include entries for individual SNPs with two key ratings. “Magnitude” is a score from 0 to 10 indicating how noteworthy the finding is. A magnitude of 0 or 1 means it’s common and unremarkable. A magnitude of 3 or above starts getting interesting. “Repute” labels the finding as “Good,” “Bad,” or neutral. A genotype page for any given SNP will show three possible genotype combinations and what each one is associated with.
These tools pull from research databases and wikis like SNPedia, where each RSID has its own page with linked studies. You can look up any RSID from your raw file on SNPedia to see what research exists for your specific genotype.
The False Positive Problem in Raw Data
Here’s something most people don’t realize: raw data files from consumer DNA companies contain a significant error rate for health-relevant variants. A study published in Genetics in Medicine found that 40% of health-related variants flagged in raw consumer data turned out to be false positives when checked with clinical-grade testing. Among those false positives, 94% were in cancer-related genes, including eight false-positive BRCA1/BRCA2 results.
This means if you’re scanning your raw data (or a third-party report based on it) and find a scary-looking variant in a cancer gene, there’s a real chance it’s a technical error. Consumer genotyping chips are designed for ancestry and common-variant analysis. They weren’t built to reliably detect rare, high-impact mutations. Any concerning health finding from raw data or a third-party tool should be confirmed through clinical genetic testing ordered by a healthcare provider before you act on it.
Getting Professional Help With Your Results
If your results flag something concerning, or if you simply want expert help making sense of complex findings, genetic counselors specialize in exactly this. The National Society of Genetic Counselors maintains a searchable directory at FindAGeneticCounselor.com where you can filter by location, specialty, and insurance participation. The American Board of Genetic Counseling offers a similar worldwide directory. Many genetic counselors now offer telehealth sessions, making access easier if none practice nearby. Your primary care doctor can also provide a referral, and most insurance plans cover genetic counseling when there’s a clinical indication.