Roadmap

Push structured facts to Wikidata and pull Wikipedia summaries, so the hub and the commons improve each other.

Every technology, target, drug, company, trial, and term has a URL and shows what links to it. That is what makes a hub rather than a list.

Verified: records something links to is 19,520/19,695 (99.1%, target 95%).

Patients, families, journalists, and investors should get the point in one sentence before the jargon.

Verified: tl;drs that are full sentences is 19,708/19,715 (100%, target 95%).

Let others build on the data: trial matchers, chatbots, dashboards. Published at /api/v1/.

An interactive force graph to navigate from a cancer to its targets to the drugs and the companies, visually.

Oncology changes weekly. Show when each fact was checked and what changed, like a package changelog.

Verified: records checked in the last 60 days is 19,715/19,715 (100%, target 90%).

Colour and badge every claim by evidence level so hype is visible at a glance.

Clinicians, scientists and leaders per institution with specialisms, roles and papers at /people/.

Verified: people with papers, or in a role that produces none (1,299 with papers, 281 not expected by role, 24 researchers still without) is 1,580/1,604 (98.5%, target 95%).

Models, companies, datasets and two roadmaps mapped as first-class objects.

Hallmarks, metastasis, microenvironment, metabolism and dormancy as pathways, terms and ideas.

Sequencing, pathology, imaging, data, manufacturing and trial infrastructure that everything else depends on.

Translate hazard ratios and pCR rates into 'out of 100 people' pictograms.

Start with Spanish, Mandarin, Hindi, Portuguese, Arabic for the TL;DR layer only, where translation is cheap and value is high.

Select your cancer type(s) and see the technologies, drugs, trials, and ideas relevant to you.

Add PD-L1, HER2-low, BRCA, TROP2 status and get a narrower, more useful view. No data leaves the browser.

Per cancer and per stage, a printable list drawn from the state-of-art and pipeline sections.

From any drug or cancer page, live recruiting trials near a postcode.

Which institutions run the key trials for this cancer; how to get referred.

Supportive care, exercise oncology, financial toxicity, fertility, and late effects deserve first-class pages.

History → current → emerging → speculative, with linked evidence. ADC, TROP2, radiopharma, cell therapy, imaging, early detection shipped.

Nightly job pulls phase 2/3 trials for every drug and target in the corpus and flags new ones for review.

Verified: products with a clinicaltrials.gov snapshot is 1,115/1,116 (99.9%, target 95%).

Expected trial readouts, FDA PDUFA dates, and advisory committees (e.g., Galleri 23 Sep 2026) on one timeline.

Within a week of each congress, update the affected objects and publish a diff.

A section for drugs and ideas that failed (TIGIT, magrolimab, rovalpituzumab, iniparib) with what was learned. Failures are data.

Combinations that work, sequences that work, and cautions, each with rationale and evidence.

Hypotheses with a proposed test, so the community can argue, refine, and eventually see them tested.

For each drug class, the known escape routes and the drugs designed to close them.

Every ADC payload/linker with permeability, efflux susceptibility, and toxicity profile, cross-referenced to ADCs.

Ac-225, Lu-177, Pb-212 production capacity and suppliers, because supply gates the radiopharma roadmap.

Where the centres that matter are, ranked by a formula anyone can inspect and dispute.

Pull Nature Index and OpenAlex counts per institution for oncology journals; publish the query.

Which institutions led the pivotal trials in the corpus. A different signal from publication counts.

SWOG, NRG, Alliance, EORTC, BIG, GBG, JCOG: what they run and how to join.

NCI, CRUK, ERC, philanthropy: where the money goes by cancer and modality.

Clinicians and scientists sign off on pages in their area; badge shows reviewer and date.

Editorial status was “shipped”, but records a named human reviewer has signed off (0 expert sign-offs, 0 advocate sign-offs; the target of 10% is 1,970 records) is 0/19,695 (0%, target 10%). Claim it.

Advocacy groups (LBBC, BCRF, TNBC Foundation, PanCAN) review TL;DRs for clarity and tone.

One named reviewer per cancer and front.

Sponsors register readouts into the calendar.

Structured summaries within a week of each meeting.

Slides and quizzes per cancer and front.

Let readers back ideas and record when an organisation picks one up.

Every object is a TypeScript record; PRs with sources are the edit mechanism. CI validates links and references.

Like the Open Medical Registry: no claim without a URL and a date. Unknown is better than guessed.

Verified: records with a primary source is 19,322/19,715 (98%, target 95%).

List missing objects and stale facts; recognise contributors who fill them.

The changelog is the newsletter. Zero extra editorial cost.

Verified organisations propose changes via /suggest/.

100 questions, a rubric, and a public leaderboard at /eval/.

Client-side index; works offline; no server.

Side-by-side of two drugs (e.g., sacituzumab vs Dato-DXd) or two technologies with the same fields.

One line of HTML to embed an OnCo object card in a blog, hospital site, or Wikipedia talk page.

Expose the corpus to AI assistants via Model Context Protocol so any chatbot can cite OnCo.

Each node links to its target page and the drugs against it.

Enter a molecular profile (mutations, IHC) and get the relevant targets, drugs, trials, and cautions in one printable view.

Curated sequences of pages: 'ADCs in 30 minutes', 'Understand your TNBC diagnosis', 'Radiopharma for investors'.

Patients bring printouts to appointments.

TNBC is first. Next: pancreatic, NSCLC, prostate, glioblastoma, each with a named expert reviewer.

Editorial status was “shipped”, but records a named human reviewer has signed off (0 expert sign-offs, 0 advocate sign-offs; the target of 10% is 1,970 records) is 0/19,695 (0%, target 10%). Claim it.

Link out to and ingest from OncoKB, CIViC, ClinicalTrials.gov, ADCdb, NCI PDQ rather than duplicating them.

Free for non-commercial and educational reuse with attribution; commercial licences fund the work.

Static export on Vercel; loads anywhere including low-bandwidth settings.

MedicalCondition, Drug, MedicalStudy markup so the hub is machine-readable to Google and AI crawlers.

Generated from the corpus: approvals, failures, roadmap progress, open problems. A yearly reference point.

Arms, N, endpoints, hazard ratios and confidence intervals on every trial.

Editorial status was “shipped”, but trials with structured outcomes, or with no results in public (1,030 with structured outcomes, 89 that have posted and published nothing, 70 with a public result not yet structured) is 1,119/1,189 (94.1%, target 95%). Claim it.

All 31 cancers now carry standard of care by setting with guideline mapping, history, pipeline, open problems, and landmark trials with structured outcomes.

Verified: cancers that reach a deep dive, their own or the parent's (322 carry their own deep dive, 149 subtype pages defer to a parent that has one, 3 reach neither) is 471/474 (99.4%, target 95%).

Percent endpoints drawn as people, time endpoints as median bars.

Route, cycle, modifications, monitoring, with the label as source.

Grade 3+ rates by event, comparable across a class at /toxicity/.

Price where disclosed, reimbursement, assistance programmes, generics.

NCCN category and ESMO-MCBS grade on standard-of-care rows where sourced.

Filings, PDUFA dates, CRLs, label changes as dated events; browse at /regulatory/.

How common each target is in each cancer, sourced; matrix at /prevalence/.

Verified: targets with sourced prevalence, or with no medicine aimed at them (207 with prevalence rows, 1,463 catalogue genes with no corpus medicine, 24 targets a medicine is aimed at still without) is 1,670/1,694 (98.6%, target 95%).

Step-by-step animated mechanism on product pages next to the molecule.

A reading level for a twelve-year-old, reviewed by advocates.

Editorial status was “shipped”, but records with a simple explanation (8,243 still to write, mostly 2,658 trials, 2,350 key papers, 1,532 targets) is 11,472/19,715 (58.2%, target 80%). Claim it.

Spanish, Mandarin, Portuguese, Hindi, marked machine-assisted until reviewed.

Editorial status was “shipped”, but records with a tl;dr in all eight languages is 7,264/19,715 (36.8%, target 50%). Claim it.

Choose questions, add your own, export one page.

Subscribe to a cancer, product, or target.

Cancer, stage, biomarkers, prior lines, country; nothing leaves the device.

What has been tried, what is next, and the cautions, at /navigator/.

Switch any page between TL;DR only and the full technical layer.

Logistics first: side effects to watch, when to call, practical support.

Glossary TL;DRs wherever a term appears.

Country and distance filters on live ClinicalTrials.gov results.

Any table state as a short URL pinned to a dashboard.

CSV, Parquet, Zenodo.

State-of-the-art panels and trial finders for hospital sites.

A PWA that caches the corpus.

Up to five items with a difference highlighter.

Cancers by targets by modalities at /pivot/.

The corpus as it stood in any year, at /timeline/.

Form-based queries over the graph at /query/.

Disclosed composite per trial and product; ranked at /evidence/.

Recomputed each build at /audit/.

A badge for clarity and tone.

Editorial status was “shipped”, but records a named human reviewer has signed off (0 expert sign-offs, 0 advocate sign-offs; the target of 10% is 1,970 records) is 0/19,695 (0%, target 10%). Claim it.

Superscripts and validation on unsourced numbers.

Last edit, author, and diff from git on every page.

Verified: records with git provenance is 19,130/19,715 (97%, target 95%).

Weekly comparison against openFDA and ClinicalTrials.gov.

Probability ranges on speculative roadmap steps and ideas.

Every factual correction at /corrections/.

Shown next to every reviewer badge.

Whether a second trial confirmed the effect, on every trial.

Select a product and see the nodes it hits and the escape routes.

A clickable body map at /body/.

Drag, zoom, hydrogens, pharmacophore colours, drug-target complexes with binding pockets.

Verified: products with a molecule or an explained placeholder is 1,074/1,116 (96.2%, target 95%).

ADC internalisation, CAR-T killing, radioligand decay, and more.

Verified: technologies with a specific schematic is 765/765 (100%, target 80%).

Light, dark, high contrast, skip link, focus rings.

Scroll-driven narrative on every roadmap.

Category matrix and animated escape-route maps per drug class at /resistance/.

Every linked object in every table explains itself on hover; headers and statuses too.

Each technology on a front page shows its schematic in motion.

Verified: technologies with a specific schematic is 765/765 (100%, target 80%).

Type scale, surfaces, tables, header, home and footer made consistent and legible in light, dark and high contrast.

45 systemic constraints, each with sourced metrics, root causes, current efforts and a page at /bottlenecks/.

Editorial status was “shipped”, but bottlenecks with ten or more ideas is 45/54 (83.3%, target 95%). Claim it.

A systematic list of what is still missing for total information dominance, turned into work.

Track each bottleneck over time: is it easing or worsening, and which ideas are being tried.

Structures, logos, OpenAlex and GLOBOCAN refreshes are manual scripts; trials, papers and the fact check already run weekly in GitHub Actions. Put the rest on a schedule too.

A pull request may not lower a health gauge that is already above its target, so coverage only moves one way.

Ten waves of 100 ideas, each with a hypothesis, a test, who must act, cost and horizon; filterable at /ideas/.

Coverage metrics computed from the graph at build time, each with the worst offenders and the fix; an idea is only 'shipped' if its gauge agrees.

Europe PMC feed and weekly trends on every drug, target, cancer and technology page, and at /papers/.

Editorial status was “shipped”, but drugs, targets, cancers and technologies with a europe pmc snapshot (368 awaiting a fetch: 174 technologies, 101 cancers, 57 targets, 36 drugs) is 3,681/4,049 (90.9%, target 95%). Claim it.

US, EU, UK, Japan, China and Australia compared product by product, gaps included.

Verified: approved products with regional rows is 453/453 (100%, target 95%).

GLOBOCAN incidence per country and cancer with data gaps made explicit at /cases/.

Output, growth, trials, burden and funders per country at /countries/.

What the leading journals, regulators and news sources are saying this month at /pulse/.