About OnCo

What it is

OnCo is a knowledge graph of 4,455 objects across 18 kinds: 64 cancers, 19 fronts, 348 technologies, 92 targets, 377 drugs, 295 companies, 421 institutions, 44 pathways, 357 terms, 401 trials, 85 pairings, 13 roadmaps, 1,107 ideas, 110 collections, 464 people, 45 bottlenecks, 146 key papers, 67 journals. Every object has its own page, a plain-English TL;DR, a technical summary, an internal last-checked date, links out to Wikipedia and primary sources, and a list of everything in the graph that connects to it. Relationships are declared once and backlinks are derived, so the graph is always consistent.

The first fully built example is triple-negative breast cancer, chosen because it went from the subtype with no targeted therapy to one with immunotherapy, PARP inhibitors, three ADCs, and a positive bispecific ADC within six years. Other cancers have state-of-the-art, standard-of-care, history, and pipeline sections at varying depth.

The rules for facts

1. Keep it current. Oncology changes weekly. Each record carries an internal last-checked date used by maintainers; the public page shows the current state, not a timestamp.
2. Prefer a link to a remembered number. Trial results quote the headline figure only when it is sourced on the page or in the linked trial record. Where a figure is approximate it says so.
3. Evidence tier is visible. Approved, phase 3, phase 2, phase 1, preclinical, concept: colour-coded on every card.
4. Ideas are labelled as ideas, with a maturity grade and a proposed test, so speculation cannot be mistaken for evidence.
5. Unknown beats guessed. Missing fields render as missing.
6. No patient data. The corpus contains public information about technologies, products, organisations, and trials only.

Institution ranking methodology

The institution table uses a disclosed formula: Newsweek points (60 minus the Newsweek/Statista World's Best Specialized Hospitals 2026 Oncology rank; zero if unranked), plus NCI designation points (Comprehensive 15, Clinical or Basic Laboratory 8), plus two points per distinct OnCo object linked to the institution. The last term rewards presence in this evidence base and is therefore also a coverage measure. The university table sums those scores by parent university and sits alongside links to Nature Index and SCImago, which are better measures of research output. Treat all of it as a starting point for argument, not a verdict.

How it is built

Next.js static export on Vercel; TypeScript data files validated by Zod schemas; a build step that checks every cross-reference resolves and emits the corpus as JSON at /api/v1/. Search runs entirely in the browser. The map uses Natural Earth country outlines from world-atlas. No server, no database, no tracking.

How to contribute

The repository is github.com/judegomila/OnCo. Each kind lives in one file under src/data/. Add or edit a record, include a source URL, run npm test (which validates the schema and every reference), and open a pull request. The Roadmap page lists what we want to build next; the failure-museum, readout calendar, and MCP server are the most requested.

Sources we lean on

FDA Oncology Center of Excellence and the AACR quarterly approval digests; ClinicalTrials.gov; NCI PDQ; NCCN and ESMO guidelines; OncoKB and CIViC; conference coverage from ASCO, ESMO, AACR, SABCS, ASH; Newsweek/Statista hospital rankings; Nature Index. See Collections for the full list with licences.

Status and disclaimer

OnCo is a work in progress. Every fact on this site is being built and checked in the open and may be incomplete, out of date, or wrong. Much of the corpus was drafted quickly from public sources and has not yet been reviewed by a named expert; pages that have been reviewed carry a badge with the reviewer and date. You must do your own research and verify anything here at its primary source (the publication, label, registry, regulator, or guideline linked from the page) before relying on it.

Not medical advice. OnCo is an orientation tool. It does not know your case. Decisions about diagnosis and treatment belong with you and your clinicians. If something here is wrong, use “Suggest an edit” on the page or open a fact correction; confirmed errors are logged at /corrections/.

Credits and data sources

Molecular structures from PubChem and the RCSB Protein Data Bank; organisation logos from Wikimedia Commons via Wikidata (licences recorded per file) with site favicons as fallback; live trial data from ClinicalTrials.gov; label checks from openFDA; research output from OpenAlex; country outlines from Natural Earth via world-atlas. Trademarks belong to their owners and are shown for identification. The editorial approach (cite everything, omit what you cannot confirm) follows the Open Medical Registry.

Licence

Code: MIT. Data (the contents of src/data/): CC BY 4.0. Attribute “OnCo (github.com/judegomila/OnCo)”.