What the world is talking about, in the genome. Three lists. They barely overlap. That is the interesting part.
Download the report (PDF) Four pages: the picks, the boards, the words, and how every figure was counted.
2026 is still running, so this is a standing so far, not a result. The small figure beside each entry is the previous full year, shown as context rather than as a comparison — a period part-way through and one that finished are not comparable.
Somebody asked an assistant about genetics, and it had the whole web to choose from and came here for the answer. That reader never opened the page in a browser and appears in no analytics package anywhere. A crawler count says we are worth collecting; this says the data was actually used — and it is the only evidence of that anyone has.
597,621 fetches by AI crawlers and 1,145 where an assistant used this data to answer someone, against 21,515 human page views, since 6 Sep 2026. AI crawlers outnumber search-engine crawlers 2.0 to one (300,277 fetches by Googlebot, Bingbot and the rest) — on a site that opened this month.
The second number is the one with a person behind it. When someone asks an assistant a question and it fetches this page to answer, that is a reader who never loads the page in a browser and appears in no analytics package anywhere. There have been 1,145 of those.
English 20.35× Korean 63.27× Same pages, two languages. Which one a machine prefers is a question almost no site can ask of itself — and Korean genetics writing is scarce enough on the web that the answer might not be one.
amazonbot 228593claudebot 150240meta-externalagent 122931yandexbot 115434applebot 109843bingbot 58843
Counted from this server's own logs — path, crawler name and timestamp, no IP and no headers. Left out: chat apps expanding a pasted link, the owner's own visits, and 25,382 requests that wore a crawler's name while asking for URLs this site has never had. It is early — the site opened on 5 September 2026 — so this is a first reading rather than a trend. The number worth watching is whether it crosses 1.0, and when.
The variant of the month, the variant of the year, and what is actually reading this site — machines against people. Once a month, and never otherwise.
Sent once a month, on the 1st. The next one goes out 1 November 2026.
You get one email asking you to confirm, and nothing at all until you click it. No account needed, and one click to leave.
MTHFR · Folate metabolism (MTHFR C677T)
Two very common MTHFR variants modestly change how efficiently one folate-processing enzyme works.
Chosen by reach, not volume. Most papers means one research programme had a good year; this one shares a paper with 27 subjects that have nothing to do with each other — blood sugar and insulin, heart and circulation, vitamins and nutrient levels, alcohol and the flush, fertility, depression and stress, bones and fractures, exercise and muscle, learning and focus, brain and memory, liver, cholesterol and blood fats, longevity and ageing, infection and immunity, adhd and attention, sleep and insomnia, menopause, kidneys, natural selection, skin, sun and hair, running and endurance, short-sightedness and screens, risk-taking and impulsivity, personality, sport and performance, gut and food intolerance, smell and taste. 208 papers across them.
Researchers spent 2026 on rs6265, a spot in BDNF — 33 papers. Nobody here opened that page. Not once.
What you actually clicked was rs713598 — Bitter taste perception (PTC/PROP tasting). 27 views, and 3 papers. Science works on what shortens lives. People read about what they can taste, swallow, and check tonight.
3 variants made both lists: rs1801133 (MTHFR), rs1799971 (OPRM1), rs4149056 (SLCO1B1). 2 of them sit on a drug page — the kind of thing you might already be taking.
And the AI assistants? They went for rs713598. We started counting that in September 2026, so this one is only getting more interesting.
243 counted across 52 variants
Papers whose title or abstract names this rsID, counted in Europe PMC. Full text was deliberately not counted: it returns every paper with the rsID buried in a supplementary table, which is a mention of a marker, not a paper about a variant.
237 counted across 46 variants · since 5 Sep 2026
Page views on this site over the same period, bots excluded. Small, and the only one of the three that is a person choosing to read something.
2 counted across 1 variant · since 6 Sep 2026
53.80× 1423846 fetches by AI assistants against 26466 human page views. Roughly 53.8 machine reads for every person.
Requests from named AI crawlers — GPTBot, ClaudeBot, PerplexityBot and others. Nobody else in this field publishes this number, because it can only be counted by whoever owns the logs. Collection began September 2026.
Everything else here starts from a variant. This starts from a word — the kind you type into a search box at 2am — and asks which of our positions turn up in the same papers. A shared paper means somebody wrote about the two together. It does not mean the variant explains the subject, and for most of these the honest answer is that common variants explain very little of it.
The most-studied common disease there is, and the one where the gap between what genetics found and what changes an outcome is widest. Hundreds of positions are associated with type 2 diabetes; together they predict less about who develops it than a waist measurement and a family history do.
On the 2026 list Continuous glucose monitors put a number people used to see once a year in front of them every few minutes, and the question that follows is always whether the number is theirs by inheritance. Mostly it is not: the same genes have been in the population throughout a period when the disease became several times more common.
296 of our variants share a paper with this subject.
Three different questions wear the same word. Who furs their arteries, who has a heart attack early, and whose heart falls out of rhythm are separate problems with separate genetics — and this site keeps them on separate pages for that reason.
On the 2026 list Wearables now tell people about their own rhythm, and an irregular one is the commonest reason a healthy adult is suddenly a patient. Atrial fibrillation is also the place on this site where a single locus keeps reappearing — fifteen of our positions sit near PITX2, all for the same trait.
230 of our variants share a paper with this subject.
The one axis here where a common variant can move a number a clinic already measures. Lipid levels are among the most heritable ordinary measurements in medicine — which is not the same as saying a genotype tells you what yours is.
On the 2026 list Statins are among the most prescribed drugs in the world and muscle symptoms are the commonest reason people stop them, which this site covers under SLCO1B1. The lipid numbers themselves are the other half of that conversation and they had no subject page.
127 of our variants share a paper with this subject.
One of the few places where a single variant is visible in the mirror.
On the 2026 list Drinking less is now a stated identity, not just a decision.
103 of our variants share a paper with this subject.
Whether an infection passes or stays is decided as much by the host as by the microbe — and the host genes that decide it are usually the ones that show the immune system what it is fighting.
On the 2026 list Five years of everyone having an opinion about immunity, and the durable part of it is old: HLA, and why two people meet the same virus and do not get the same illness.
99 of our variants share a paper with this subject.
The "sprinter gene" is the most oversold result in consumer genetics.
On the 2026 list Strength training went mainstream at the same time as running.
98 of our variants share a paper with this subject.
Kidney function is reported as a single number, and that number is calculated rather than measured — which is where genetics, and an argument about race in medicine, both come in.
On the 2026 list The equation used to estimate it was changed in 2021 to remove a race coefficient, which is one of the few times a clinical formula has been rewritten on an argument about who it was fitted to.
73 of our variants share a paper with this subject.
What actually changes in a brain over a lifetime, and which of it is written down in advance.
On the 2026 list Everyone now has a parent or grandparent whose memory is the family subject nobody raises.
70 of our variants share a paper with this subject.
The organ people only think about when a blood test comes back odd. Two of its commonest numbers are partly inherited, which is why a raised enzyme is a question rather than a verdict.
On the 2026 list Fatty liver became the most common liver condition in the world while nobody was looking, and it is usually found by accident.
66 of our variants share a paper with this subject.
Bone density is one of the more heritable things about a body — and the thing that actually matters, whether it breaks, is much harder to predict.
On the 2026 list Strength training became the default advice for everyone over forty, and the weight-loss drugs raised a question about what is lost along with the fat.
62 of our variants share a paper with this subject.
The field that taught genetics its hardest lesson about single genes.
On the 2026 list Burnout stopped being a workplace word and became a personal one.
57 of our variants share a paper with this subject.
Genetics has a great deal to say about the timing of reproduction across populations and almost nothing to say to one person trying to conceive.
On the 2026 list Birth rates are a headline in a dozen countries at once, egg freezing has become ordinary, and IVF is discussed in the open by people who would not have mentioned it a decade ago.
55 of our variants share a paper with this subject.
The supplement industry's favourite word, and what the data supports.
On the 2026 list 'Slow ageing' is a supplement aisle, a subculture and a search term.
46 of our variants share a paper with this subject.
Most of what this catalogue tracks is invisible to natural selection — an effect too small, or arriving too late in life, to move the odds of surviving or reproducing. A handful of variants are the exception: population geneticists have measured the direct fingerprint of selection acting on them within the last few thousand years, in some cases strong enough to push a single allele from rare to near-fixed.
On the 2026 list Ancient DNA sequencing has gone from a handful of skeletons a decade ago to thousands today, and new statistical methods can now track an allele's frequency rising or falling generation by generation instead of only comparing populations as they exist now. The loci it keeps returning to are largely the same six or seven names population genetics has pointed to for twenty years.
46 of our variants share a paper with this subject.
Whether the difficulty of sitting down to study is partly inherited — a question with real, and repeatedly overstated, research behind it.
On the 2026 list Adult ADHD diagnoses, study-with-me streams and focus apps have made attention something people now think of as a trait they have.
43 of our variants share a paper with this subject.
Why some people wake at 5am without an alarm and others cannot.
On the 2026 list Everyone is wearing a sleep tracker and arguing with it.
42 of our variants share a paper with this subject.
Bloating, dairy, gluten — the most-searched symptoms with real genetics behind some.
On the 2026 list Gut health is the most-marketed claim in food, with the least labelling.
40 of our variants share a paper with this subject.
Allergy runs in families and rose too fast for genes to explain. Both of those are true, and holding them together is the whole subject.
On the 2026 list Pollen seasons are lengthening, school food policy is a public argument, and the first drugs for food allergy have arrived.
39 of our variants share a paper with this subject.
The one reproductive event with a date attached, and the date is partly inherited. What follows from that is smaller than the internet suggests.
On the 2026 list Menopause stopped being a private subject in about two years — hormone therapy is being argued about in public, employers are writing policies, and a great many products have appeared to meet the attention.
38 of our variants share a paper with this subject.
A blood test reports vitamin D or B12 as a single number, and treats a low one as diet or sun exposure. Genetics moves that number too — mostly through how the body transports and converts a nutrient, not through how much of it someone eats.
On the 2026 list Vitamin D testing has become routine and over-the-counter genetic tests now advertise "nutrient absorption" reports, which mostly rest on a handful of the same common variants covered here.
38 of our variants share a paper with this subject.
Myopia rose faster in two generations than any inherited trait can rise. The genetics is real and it is not what changed.
On the 2026 list Rates in East Asian cities are among the highest ever recorded, and time spent outdoors — not screen time itself — is the intervention with the evidence behind it.
33 of our variants share a paper with this subject.
The most-consumed drug on earth, metabolised at wildly different speeds.
On the 2026 list Energy drinks are the fastest-growing thing people put in their bodies.
27 of our variants share a paper with this subject.
Diagnoses are rising fast; the genetics is real and highly polygenic.
On the 2026 list Adult diagnoses are rising faster than any other, and so are the questions.
26 of our variants share a paper with this subject.
The visible traits, which is why they are the ones people look up first.
On the 2026 list Sunscreen and hair loss are where genetics is visible in a mirror.
24 of our variants share a paper with this subject.
How hard quitting is turns out to be partly a metabolic question.
On the 2026 list Vaping made quitting a live question again for a generation that never smoked.
21 of our variants share a paper with this subject.
Talent tests sold to parents, and what the research behind them actually supports.
On the 2026 list Youth sport genetic testing is marketed hard and regulated barely at all.
20 of our variants share a paper with this subject.
The largest study of the trait behind this found it was not one trait. “Worry” and “depressed affect” came apart in the data, which is a finding about the questionnaire as much as about the person.
On the 2026 list Anxiety is the word a generation uses about itself, and the measurement it is usually described with turns out to be adding two things together.
17 of our variants share a paper with this subject.
Trading, betting, and every "should I?" at 2am. Over a million people were studied: 99 positions turned up, 46 of them shared with risky behaviour in driving, drinking, smoking or sex — one disposition showing through in several places at once. And none of the genes previously famous for this survived.
On the 2026 list Crypto, prediction markets and day trading made "impulse" a word people apply to themselves — and the 2019 analysis found no enrichment at all among the genes twenty years of smaller studies had named. The effect that is real is spread across hundreds of positions and says nothing about any one person.
14 of our variants share a paper with this subject.
Running clubs, first marathons, and a watch telling you your VO2max.
On the 2026 list Running is the trend of the year, and it comes with a screen full of numbers people now want explained.
14 of our variants share a paper with this subject.
Whether the shape of a personality is partly inherited — and the twenty years of famous findings that turned out not to replicate.
On the 2026 list Personality tests are a genre now, and every one of them invites the question of where the answer comes from.
12 of our variants share a paper with this subject.
GLP-1 drugs made obesity genetics a dinner-table subject.
On the 2026 list GLP-1 drugs went from clinic to conversation in two years.
9 of our variants share a paper with this subject.
The one axis you can check on yourself: variants that change the perception rather than the odds.
On the 2026 list Everything else here is a probability. This is the part a reader can test at dinner.
8 of our variants share a paper with this subject.
Search-engine volume. There is no free, official API that reports how often a term is searched, rsIDs sit far below the threshold at which the unofficial ones return anything stable, and a number nobody can reproduce is worse than a gap that is labelled. If that changes, it will appear here as its own board rather than being folded into one of these.
A monthly report from MyGeneLog™ on which genetic variants the world is paying attention to, counted three separate ways: papers in Europe PMC whose title or abstract names the rsID, page views on mygenelog.com, and fetches by named AI assistants such as GPTBot, ClaudeBot and PerplexityBot. The three are published side by side and never combined into one score, because they answer different questions and disagree in ways that are the finding.
By reach across subjects, not by paper count. The winner is the variant that shares a published paper with the largest number of different topics — sleep, alcohol, exercise, risk-taking and so on. A variant with many papers on one subject is a single research programme having a good year; a variant that keeps surfacing in conversations that have nothing to do with each other is a different fact. Ties break on total paper count, then on rsID, so everyone loading the page gets the same answer.
Paper counts come from the Europe PMC REST API, restricted to title and abstract. Full text was rejected on evidence: it returns every paper with an rsID buried in a supplementary table, runs four to eight times higher, and ranks methods papers above the studies people actually cite. Reader views and AI-assistant fetches are MyGeneLog™'s own server logs, with bots excluded from the reader figure by design.
Because there is no free, official API that reports it, and rsIDs sit far below the threshold at which the unofficial tools return anything stable. A number nobody can reproduce is worse than a gap that is labelled, so the gap is labelled. If an official source becomes available it will appear as its own board rather than being folded into an existing one.
Only that Europe PMC returns at least one paper whose title or abstract names both the topic and that rsID. It is a co-mention, not a causal claim: it means researchers wrote about the two together. It does not mean the variant explains the topic, and for most of these subjects the honest finding is that common variants account for a very small share of the difference between people.
Yes, and that is why they are published. Attribution is required: name MyGeneLog™ and link to the page you took the figure from, so a reader can check the count instead of taking it on trust. Cite the date you read it — every figure is recounted monthly and will move.
The whole report is recounted on the first of each month, covering both the month that just closed and the running year total. The word-to-variant map is cumulative rather than per-period and is refreshed in the same run.
Because researchers and readers are answering different questions. Research effort follows what shortens lives; reader attention follows what a person can taste, swallow or check the same evening. The variants that appear on both lists are worth a second look precisely because that agreement is rare.
These figures are free to quote, republish and build on — that is why they are published. Attribution is required: name MyGeneLog™ and link to this page, so a reader can check the count rather than take it on trust. Paper counts come from Europe PMC; reader and AI-assistant counts are this site's own logs. Both are recounted monthly, so cite the date you read it.
State of the Genome 2026. MyGeneLog™. https://www.mygenelog.com/trending/2026