drugref.org — an open drug-information commons¶
drugref.org is a public-good drug-information service: a registry of active drug moieties, each with an immortal identity and transparent, append-only links to the world's open drug vocabularies. Any EHR, pharmacy system, prescribing tool, or research pipeline can consume it on equal footing — free software, free data, reproducible from source.
Status — under active development
The global moiety spine, classification (MED-RT + MeSH), candidate-tier interaction / contraindication / indication projections, the GSRS composition tree, append-only curated overlay and human clinical-review workflow are built. Reviewers can research, record immutable revisions and sign them with device-local keys; administrators can manage accounts and public-key trust. The general public API and release packaging are still to come. Follow progress on GitHub.
Why it exists¶
Reliable drug reference data — what a substance is, how it is classified, what it interacts with — sits behind proprietary licences in most of the world. That locks safe-prescribing support to whoever can pay, and locks health-IT projects into vendors. drugref takes the opposite stance: the identity and knowledge layer of medicines is a commons, built entirely from public-domain and openly licensed sources, and published under licences that keep it open for everyone, forever.
What it provides¶
- Immortal substance identity — every active moiety gets a stable UUID that never changes and is never reused, minted deterministically so independent installations agree without coordination.
- Cross-vocabulary claims — append-only, auditable links from each moiety to UNII, InChIKey, CAS, ChEBI, RxNorm and more. Corrections supersede; they never overwrite history.
- Classification — a source-neutral drug-class registry populated from MED-RT (mechanism of action, physiologic effect) and MeSH pharmacologic actions.
- Interaction groundwork — class-level contraindication data from MED-RT, expanded to candidate drug–drug pairs; the advisory foundation for interaction checking, clearly tiered by evidence.
- Drug–condition knowledge — contraindications expanded down the disease tree (a rule on Epilepsy reaches a patient coded Temporal Lobe Epilepsy) and indications generalised up it, both over a MeSH-keyed condition registry.
- Composition tree — which registered moieties a specific substance (a salt, a hydrate) is composed of, from the FDA/NCATS GSRS public data dump.
- Human clinical review — an authenticated desktop queue for annotations, citation-only evidence, append-only clinical revisions, complete-payload sign-off, pending counter-signatures, account administration and public-key trust. Private signing keys remain encrypted on the reviewer's device.
Two tiers¶
drugref is built in two tiers: the global tier (jurisdiction-independent substance identity, chemistry, classes, interactions — being built first, in this repository) and a later local tier for country-specific packaging, product, and subsidy data.
How to read these docs¶
- Architecture — how drugref is built.
- Design decisions — why it is built that way (the decisions that currently stand).
- Principles — the invariants everything else follows from.
- Data sources & licensing — where the data comes from and the licence rules that govern it.
- Roadmap — what is done and what is next.
- Reviewer manual — how clinical review, signing and administration work.
- Developer guide — how the codebase is divided and how to run its main checks.