Each page covers one condition properly: what it actually is and what it feels like, what causes it, how common it really is and in whom, and how it's diagnosed — followed by the specific variants with a replicated link to it, and, where the research supports it, how genetics can change the way a medicine behaves.
Written in two registers so one page serves both readers: plain language first, then a clinical detail section with the precise terminology, effect sizes and diagnostic thresholds. Every claim is sourced from freely-available published research and cited on the page, so you can check us — or hand the page to your doctor. 385 conditions so far, growing continuously, and free to reuse with attribution (CC BY 4.0).
One gene, more than two thousand described variants, and a treatment that is chosen by which variant somebody carries rather than by the diagnosis they have. It is the clearest case anywhere of why the variant is the unit that matters.
Read more → PharmacogenomicThe enzyme where the broken copy is the common one. Most people of European ancestry carry two of them, which is why the standard dose of one transplant medicine is the dose that suits a non-working enzyme — and why the people whose enzyme works are the ones who get too little.
Read more → PharmacogenomicOne dose of an ordinary antibiotic, at an ordinary blood level, and the hearing does not come back. It happens to people carrying one change in mitochondrial DNA — which their mother also has, and so do all of her children.
Read more → PharmacogenomicThe most common enzyme deficiency in humans, and for almost everyone who has it nothing ever happens. It matters on the days it does: a short list of medicines, and a bean.
Read more → PharmacogenomicA person who is perfectly well until they are given one particular kind of anaesthetic, and then is not. It is the clearest case in medicine of a genotype that changes a decision — and the decision is simply to use a different drug.
Read more → PharmacogenomicOne enzyme, four prescribing guidelines. CYP2C19 decides how much drug reaches you for a heart medicine, a stomach medicine, an antifungal and most antidepressants — and the same genotype helps with one and hinders another.
Read more → PharmacogenomicThe clearest result in all of pharmacogenomics: test before the first dose, do not prescribe if positive, and a reaction that used to be feared largely stopped happening. It is also the same allele that makes some people control HIV unusually well.
Read more → DevelopmentalOne of the most common birth differences in the world, and the locus that carries the most risk has two candidate causal variants — with the evidence for which one matters depending on which population you ask.
Read more → AutoimmuneAutoimmune thyroid disease affects 2–5% of people, and the antibody that marks it is often present years before anything goes wrong. The genetics of having the antibody is not the same question as the genetics of getting ill.
Read more → PharmacogenomicCotinine is the test used to check whether somebody smokes. It turns out how fast you clear it is partly genetic — which matters because smokers of different ancestries get lung cancer at different rates even after adjusting for how much they smoke.
Read more → NeurologicalOne of the most heritable conditions in psychiatry, and its first genome-wide study found nothing that met the threshold. The variant on this page is that study’s top signal — published because a near miss in a small sample is information too.
Read more → CardiovascularA genome-wide scan of venous thrombosis found no new genes at all. That is the result worth reporting: the four loci already known account for what common variants explain, and one of them is your blood group.
Read more → InfectiousWhether an infection becomes chronic is decided largely by the host, and the strongest host signal is in the genes that decide what the immune system is shown.
Read more → Trait449,484 people, 136 positions, 599 genes — and the most interesting result is that the trait split in two. What one questionnaire measures as one thing turned out to be two genetically distinguishable things.
Read more → MusculoskeletalThe study behind this page did something unusual and useful: it looked for genes for the fracture rather than genes for the bone density. It found one, at an odds ratio of 2.25.
Read more → NeurologicalNine in ten cases have no family history, and for a long time that ninety percent was the part genetics could say least about. It took 13,225 people to find one new position.
Read more → NeurologicalA rare, rapidly fatal prion disease of the brain — a 2020 GWAS, the first for this condition, found a genome-wide significant risk variant near STX6, a gene already separately implicated in another neurodegenerative disease.
Read more → CancerThe most common cancer in the world, and one of the least discussed. Its genetics was found by sequencing 2,636 Icelanders and imputing into a quarter of a million people — and one of the variants sits in TP53 itself.
Read more → AutoimmuneTwenty-two positions replicated, and every one of them was in the MHC. What makes it interesting is that they were not in strong linkage with each other — this is not one signal, it is several, stacked in one region.
Read more → CardiovascularThe largest meta-analysis of stroke genetics found that the loci belong to subtypes rather than to stroke. A clot from the heart and a blocked neck artery are different diseases that end in the same emergency.
Read more →Showing 20 of 385 · page 16 of 20
One diagnosed condition or measurable trait covered properly: what it actually is, what causes it, how common it is, how it is diagnosed, and the specific variants with a replicated link to it — every claim sourced from published research.
No. Every page here is informational, not a diagnosis. A result should be discussed with a doctor or a licensed genetic counselor, never acted on alone.
Because they are not disorders. Height, ABO blood group and a handful of others describe a measurable characteristic with nothing to diagnose — "Trait" says so plainly instead of implying a disease that is not one.
Only after a replicated association in published research, verified against the primary study rather than a gene name or a marketing claim — the same discipline the variant pages themselves follow.