The data

What connects to what.

Finding a treatment for a rare disease usually means noticing that something already known about one gene, one drug or one pathway applies somewhere nobody has looked. That only works if the connections are there to follow. These are the connections we hold — drawn by kind, because the kinds are the part worth seeing at a glance.

Every circle below is a category, not a thing: “gene”, not CASK. Every line is a count of how often that kind of link occurs. The individual links are not published here.

Read from the literature

What our panel drew out of the papers

When the panel reads a paper it records the claims that paper makes — this gene causes this disease, this drug treats this condition — each one attached to the paper that made it. This is the shape of everything it has recorded.

10,421 relations from 764 papers. A line is a count of claims of that kind, and its direction is the direction the claim was made in — genes act on diseases far more often than the reverse, which is what the literature says rather than something we decided. The 12 heaviest shapes are drawn; 20 rarer ones are not, and they are the thin tail rather than a hidden category.

A claim recorded is not a claim endorsed. Each of these is something a paper asserted, kept with the paper that asserted it and with the panel’s reasoning about it. Recording it is how the assertion becomes checkable — not a statement that it is true.

Assembled reference

And what the substrate can traverse

Underneath sits a much larger reference map, assembled from published biomedical databases. It is not our work and it is not what we publish — it is what the engine is allowed to walk when it looks for a connection between two things nobody has studied together.

Types and counts only. These relationships come from licensed third-party sources, are used in the backend to inform our own conclusions, and are never returned to anyone — which is why this page can say how many links of each kind exist, and not which. The 10 heaviest shapes are drawn; 16 rarer ones are not, and they are the thin tail rather than a hidden category.

The two graphs are different kinds of thing, and the larger one is not the achievement. The map above is a reference anyone can license. The one before it is what a three-model panel read, argued about and wrote down, and it is the part that did not exist until we made it.

What this is not

A shape, not an answer

A connection is a place to look, not a finding. That a gene and a disease are linked in the literature says someone reported it. Whether it holds up is the work that comes after, and most of it is not done.
Thickness is frequency, not strength. A heavy line means that kind of claim is made often. Claims made often are sometimes just claims that are easy to make.
The underlying links are not published. Not on this page and not anywhere on this site — the licensed ones for licence reasons, and the rest because what we offer is the analysis, not the source material behind it.

Our position What we hold evidence on Request access

Research aid only — not medical advice. Piper is a research and educational aid. It is not medical advice, not a diagnosis, and not a clinical determination. Its output is hypothesis-generating and may be incomplete, provisional, or wrong, including AI-generated errors. Always consult your own qualified healthcare professional before making any medical or treatment decision.

The material on this site is written with the help of AI and may contain mistakes. It describes the state of published knowledge rather than any individual, and every source it rests on is named so you can check it.