Disease Intelligence. Connected.
Open-source research software for connecting disease knowledge, evidence, and provenance across biomedical sources.
Explore the documentation · Run locally · Contribute
Public alpha · research use only. NosoGraph is not medical advice, a diagnostic system, or clinical decision support. No public hosted demo is deployed; use the local Docker evaluation or CLI.
| Capability | What it provides | Maturity |
|---|---|---|
| Explore evidence | Follow disease context through typed claims, evidence direction, source context, and provenance | Public Alpha |
| Compare conditions | Compare two to five conditions with explicit missingness and deterministic JSON/Markdown exports | Beta |
| Build research workflows | Use the CLI, FastAPI surface, disease modules, and 40+ analysis pipelines locally | Stable CLI · Beta API |
| Extend the graph | Add disease curation, source adapters, validation, documentation, and analysis code | Open source |
NosoGraph complements upstream resources such as MONDO, HPO, PubMed, ClinicalTrials.gov, Open Targets, and GWAS Catalog. It does not replace them or turn associations into causal conclusions.
- Understand NosoGraph
- Follow the five-minute SLE workflow
- Inspect claims and provenance
- Compare conditions
- Review source coverage and limitations
- Install from source
- Understand the architecture
- Use the CLI or API
- Run the test suite
- Contribute code, curation, sources, or documentation
git clone https://github.com/AdamEddahmouni/nosograph.git
cd nosograph
cp .env.example .env
docker compose --profile full up --buildOpen http://localhost:8000. Docker Compose v2 starts the API, worker, and Redis. For a CLI-only or contributor installation, follow the installation guide.
The product CLI is nosograph. The installable distribution and import path remain med-research and med_research during the public-alpha compatibility period.
Disease → Typed claim → Evidence relationship → Study / source → Provenance / snapshot
Evidence records support claims; they are not automatic proof. Supporting, contradictory, inconclusive, and unasserted evidence can coexist. Missing metadata remains unknown rather than silently becoming certainty. See the evidence model, provenance model, and data-source matrix.
NosoGraph v0.2.1 · Public Alpha · repository snapshot 2026-08-22
| Repository-backed measure | Value |
|---|---|
| Registry modules | 10,407 |
| Strict L2-validated modules | 88 |
| Reference modules | 6 |
| CI-validated modules | 8 |
| Offline tests selected in the v0.2.1 suite | 2,445 |
Values come from docs/generated/public-status.yaml. Registry breadth is not curation depth: most registry modules are scaffolds. See the current project status for capability-level maturity and limitations.
NosoGraph welcomes focused contributions to code, documentation, source integration, and disease curation. Start with CONTRIBUTING.md, then run the local gate:
make ci-localQuestions belong in GitHub Discussions. Report vulnerabilities privately through SECURITY.md. Never submit secrets, PHI, or patient-identifiable data.
Cite v0.2.1 with the all-versions concept DOI 10.5281/zenodo.22055279 until a v0.2.1 archive record exists. The historical version DOI 10.5281/zenodo.22062925 identifies v0.2.0 only. Canonical metadata is available through GitHub's citation UI and CITATION.cff.
@software{nosograph2026,
title = {NosoGraph: Disease Intelligence. Connected.},
author = {Eddahmouni, Adam and NosoGraph contributors},
year = {2026},
url = {https://github.com/AdamEddahmouni/nosograph},
version = {0.2.1},
doi = {10.5281/zenodo.22055279}
}NosoGraph source code is available under Apache-2.0. Upstream biomedical datasets retain their own terms; see data licenses. Outputs are computational research artifacts—not diagnoses, treatment recommendations, or clinical decision support.