I re-test almost everything. Blood work quarterly. [[Epigenetic clocks::Epigenetic clocks: estimates of biological age read from chemical marks on your DNA that shift with age and lifestyle, separate from the fixed genome underneath.]] quarterly. Body composition, [[VO2 max::VO2 max: the maximum rate of oxygen your body can use during hard exercise, one of the tightest fitness predictors of longevity.]], sleep, recovery, all on some loop that resets and asks the question again. Every number I track about my body is a snapshot with a short shelf life, because the body it measures keeps moving. My genome is the exception. I read it once. It will be as true on the day I die as it was the day I spat into the tube. Of all the data I collect chasing a longer life, the DNA is the only piece that never expires, and it’s the one most people never bother to read. I’m 31. I’ve had my genome [[genotyped::Genotyping: reading specific pre-chosen positions in your DNA on a chip, as opposed to sequencing, which reads the full genome letter by letter.]] for years. Here’s what it actually told me, what it ruled out, and where it goes quiet. The first thing 23andMe shows you is the fun part. My composition comes back 100% European, and inside that, 94.4% French, with a Paris Basin match specific enough to feel like it’s reading my mail. A 5% seam of Dutch and Northern German, thin traces of Greek and Scottish under a percent each. Nothing surprising to me, but seeing it rendered as a number has its own gravity. Then it goes deeper in time. My maternal line is [[haplogroup::Haplogroup: a population sharing a common ancestor on a direct maternal or paternal line, traced through mitochondrial DNA or the Y chromosome.]] J1c3, which walked into Europe from the Near East with the first farmers about nine thousand years ago and traces to the same maternal ancestry that identified Richard III under a Leicester car park in 2013. My paternal line, R-Y3444, is a rare branch of the R-M269 lineage the Bronze Age steppe migrations carried across Europe, and it runs to the House of Bourbon. Another 254 stretches of my DNA come from Neanderthals, more than seventy percent of the people 23andMe has tested, though still under two percent of the whole. That is the candy. It sells the kits, it makes for good dinner-party trivia, and the relative-matching behind it is the same mechanism that let investigators catch the Golden State Killer through a distant cousin’s DNA. For your health, though, ancestry barely matters. Where you came from is a story about the past, and the part of the genome actually worth acting on shows up in the health screen, which is where the rest of this goes. Most of a genetic health screen is a wall of the same three words: variant not detected. Mine reads that way almost all the way down. No [[BRCA1 or BRCA2::BRCA1 and BRCA2: genes whose harmful variants sharply raise the risk of breast, ovarian, and some other cancers.]] among the 44 variants tested, including the founder mutations that run through Ashkenazi families. Zero of 46 [[carrier variants for recessive conditions::Carrier status: having one copy of a recessive-disease variant, which does not affect you but can affect a child if the other parent carries it too.]]. No [[Factor V Leiden::Factor V Leiden: an inherited mutation that raises the risk of abnormal blood clots.]], no [[familial hypercholesterolemia::Familial hypercholesterolemia: an inherited disorder causing very high LDL cholesterol from birth and early heart-disease risk.]] across a two-dozen-variant panel, no [[hemochromatosis::Hemochromatosis: an inherited tendency to absorb and store too much iron, which can slowly damage organs.]], no [[G6PD deficiency::G6PD deficiency: an inherited enzyme deficiency that can trigger red-blood-cell breakdown after certain foods, drugs, or infections.]], no Parkinson’s-linked variant. And on the single strongest common genetic risk factor for late-onset Alzheimer’s, the [[APOE gene::APOE: a gene whose ε4 variant is the strongest common genetic risk factor for late-onset Alzheimer’s disease.]], I came back ε4-negative. None of that is a clean bill of health, and I’ll get to why. But ruling things out is not a consolation prize. It’s most of the point. The genome doesn’t usually tell you what will happen to you. It tells you where not to look, which of the thousand things you could reasonably worry about actually deserve a slice of your attention. The familial hypercholesterolemia result is the clearest example. I watch my lipids. Knowing there’s no monogenic driver behind them means any number I see has a lifestyle and context explanation, not a broken gene I need to chase. That changes what I do with the reading. There was one flag. Two copies of a common variant in the [[CFH gene::CFH: the complement factor H gene, whose variants are among the largest common genetic contributors to age-related macular degeneration.]], one of the largest-effect common variants known for [[age-related macular degeneration::Age-related macular degeneration: a leading cause of vision loss in older adults, affecting the central field of view.]]. 23andMe files it as a slightly increased risk, since the disease is age-driven and decades away, but the flag is specific and real. It’s not destiny. What it is, is a standing note: when I’m older, take the eye exams seriously, and the interventions that help are the same ones already in the longevity playbook. One line in the whole report changed a future decision, which is a fair return on a test I only had to take once. It’s easy to oversell this. 23andMe doesn’t sequence your genome. It genotypes it, checking specific pre-chosen positions on a chip rather than reading all three billion letters. For a given condition it’s testing a handful of common variants, so a “variant not detected” means exactly that and nothing broader. My BRCA result clears the 44 spots it looked at, not the thousands of rarer mutations a full clinical sequence would catch. And because that panel leans on mutations common in Ashkenazi families, a clean result carries less weight for my mostly-French DNA than the same result would for someone whose ancestry matches the test. A lot of the reports are [[polygenic::Polygenic: shaped by many genes each with a tiny effect, summed into a risk score rather than a single yes-or-no variant.]] anyway, scores summed across many small-effect variants rather than a single verdict. Those are population statistics wearing the costume of a personal result. And the one category I’d actually want for daily decisions, [[pharmacogenetics::Pharmacogenetics: how your genes affect the way your body processes specific drugs, used to guide drug choice and dosing.]], how my body processes specific drugs, sits behind a premium upgrade I don’t have, so the reports most likely to change an actual prescription are the ones I can’t see. So the read is real but shallow. It’s a wide, thin pass over the genome, good at ruling out the well-known bad actors and honest about very little else. Treating it like a clinical sequence is the mistake that makes people either panic or relax for the wrong reasons. There’s a question you have to sit with before you spit in any tube, and the longevity crowd tends to skip it. Your genome is the most permanent identifier you own. You can change a password after a breach; you can’t change your DNA. It can’t truly be [[de-identified::De-identified: stripped of obvious personal identifiers, though a genome is close to unique and can often be re-linked to a person through public genealogy databases.]] either, because you share big stretches of it with relatives who never signed up for anything. Roughly 60% of Americans of European descent can already be identified from a distant cousin in a consumer database, whether or not they ever tested. None of this is hypothetical. In 2023, attackers [[credential-stuffed::Credential stuffing: breaking into accounts with username and password pairs leaked from other sites, exploiting people who reuse passwords.]] their way into 23andMe and left with profile and ancestry data on roughly 6.9 million people. What they did next is the part worth remembering: they compiled and sold lists that specifically singled out users with Ashkenazi Jewish and Chinese ancestry. A UK regulator later fined the company £2.3 million for the security failures behind it. The data was collected to tell people about their health and their ancestry, and it got turned into target lists by ethnicity. The everyday version of the tradeoff is quieter. Your genome is not just stored, it’s the product. 23andMe’s real business was selling access to the aggregate database, including a $300 million partnership with a pharmaceutical company, with around 80% of customers opted into research. And the law only half-covers you: [[GINA::GINA: the Genetic Information Nondiscrimination Act of 2008.]] bars genetic discrimination in health insurance and employment, but says nothing about life, disability, or long-term-care insurance, which can still ask. Then there’s what happens when the company changes hands. In 2025 23andMe filed for bankruptcy, and fifteen million people’s genetic data became an asset in a court proceeding. A drug company won the first auction; a nonprofit founded by the former CEO won the reopened one, with privacy pledges attached. A court-appointed privacy watchdog concluded most customers never understood they had agreed to any of this. Consent you give one company does not bind whoever owns it next. The read is still worth having. What I’m less sure of is whether a consumer platform, whose incentives can change hands overnight, is the right place to keep it. The raw file is the part worth owning outright, and the terms are worth reading before you spit. Here’s the frame that makes the genome worth the trouble. It’s the hand you were dealt, and it does not change. Everything else I measure, the epigenetic clocks, the blood panels, the organ-level aging estimates, is a readout of how I’m playing that hand right now, and all of it moves. My biological age can drift years between two draws of the same blood. My genome gives the same answer every time because it is the same thing every time. That permanence is the whole argument for reading it early and once. The epigenetic layer is where the action is, where diet and training and sleep actually show up, and it’s noisy and humbling in equal measure. But the fixed hand underneath sets the priors. It tells you which flags in the moving data to take seriously and which to let pass. My CFH variant means an eye metric years from now isn’t noise. My clean lipid genetics means an [[LDL::LDL: low-density lipoprotein, the cholesterol fraction most tied to cardiovascular risk.]] number is a lever, not a warning. Reading it once also leaves you a library you can keep querying. A few weeks ago I fell down a rabbit hole on coffee and kept hitting [[CYP1A2::CYP1A2: the liver-enzyme gene that sorts people into fast and slow caffeine metabolizers, based on a single common variant.]], the gene that splits people into fast and slow caffeine metabolizers. I didn’t have to test anything. My genotype was already indexed and searchable in the same 23andMe data I gave years ago, one lookup away from a question I didn’t have back then. The sequencing itself gets cheaper every year, and whole-genome reads that cover nearly every letter already sell for a few hundred dollars. When that fully arrives the read gets deeper, but the logic holds: read the fixed part once, early, keep it somewhere you control, and spend the quarterly budget on the parts that move. The hand you’re dealt is also the one you can’t hand back, so read it with both facts in view. The genome won’t tell you how you’re aging. It tells you where to point the instruments that do.
