Open-source biomedical research software

Trace disease evidence from claim to source.

NosoGraph connects disease knowledge from MONDO, HPO, literature, and clinical trials into typed claims — and keeps every claim's evidence direction and source provenance open to inspection.

Start with the docs Run NosoGraph locally View source →

This site is the documentation · the software runs locally · no hosted demo

Evidence trace Illustrative structure

1. 01DiseaseMONDO:0007915
2. 02Typed claimassociated_with
3. 03EvidenceSUPPORTS
4. 04Study / sourceNOT_RECORDED
5. 05Provenancesnapshot context

Disease context systemic lupus erythematosus MONDO:0007915 · reference disease module

Repository-backed vocabularyUnknown stays unknown

Follow a typed claim through evidence, source context, and preserved provenance.

Research software — not medical advice Status: public alpha · v0.2.1

Current scope

A broad registry, honestly labeled.

Full project status →

Discoverable disease modules

10,404

full-corpus

CI-validated modules

8

full-corpus gate

Reference modules

6

full-corpus

Strict L2-validated

88

n=500 sample

Registered pipeline adapters

25

full-corpus

Offline tests

2,445

v0.2.1 snapshot

Registry breadth is not curation depth. Validation figures marked sample describe an n=500 alphabetical slice of the registry, not the full corpus. How the registry counts work →

Positioning

Research software, with the boundary on the page.

NosoGraph is

• Open-source (Apache-2.0) software for biomedical research workflows.
• A typed-claims layer across ontologies, literature, trials, genetics, and target resources.
• Local software you run and inspect — a CLI, an API, and a dashboard.
• Provenance-preserving: source, snapshot, and import context travel with each claim.

NosoGraph is not

• Not medical advice, diagnosis, or clinical decision support.
• Not a hosted service — this site is documentation; the app runs on your machine.
• Not a replacement for the upstream sources it reconciles.
• Not a truth engine — computational rankings are hypotheses for review, never clinical certainty.

Choose your path

Inspect the evidence or extend the system.

Major capabilities

What you can inspect today.

Each surface is documented with its actual maturity. Nothing here is a clinical result.

Public alpha

Evidence Explorer

Inspect condition context, typed claims, and evidence direction in the local dashboard.

Read the guide → Beta

Evidence Workspace

Turn selected evidence sources into a provenance-backed dossier, asynchronously.

Run it locally → Beta

Compare conditions

Two to five conditions, explicit missingness, deterministic structured output.

Read the workflow → Stable

Disease registry

10,404 discoverable modules — breadth labeled separately from curation depth.

Understand the registry → Stable

CLI

Validate disease modules, sync sources, and run task-oriented analysis commands.

CLI guide → Beta

API

Local FastAPI surface for diseases, evidence, provenance, and comparison.

API reference →

The NosoGraph layer

From fragmented sources to inspectable relationships.

Biomedical resources are built for different questions. NosoGraph adds a computational layer across them — not a replacement — and keeps the distinction between source data, typed relationships, evidence, and provenance visible at every step.

1. 01Source syncAcquire a resource record and retain its snapshot context.
2. 02NormalizeResolve identifiers while retaining source vocabulary.
3. 03Typed claimsRepresent explicit subject–predicate–object relationships.
4. 04EvidenceKeep SUPPORTS, CONTRADICTS, and INCONCLUSIVE distinct.
5. 05ProvenanceCarry source, snapshot, import, and fingerprint context forward.

Evidence is direction, context, and provenance — not a single score.

NosoGraph structures evidence for inspection. It does not automatically turn a claim into scientific proof or a clinical conclusion.

ASSOCIATED_WITH is a relationship type. It is not automatically CAUSES.

Evidence semanticsStructural example

Typed claimsubject → predicate → objectassociated_with

SUPPORTS

Evidence supports the claim; it is not a proven fact.

CONTRADICTS

Evidence disagrees with the claim; it is not automatic falsification.

INCONCLUSIVE

Evidence does not establish a clear direction.

UNASSERTED

No directional evidence is recorded.

Missing fieldNOT_RECORDEDProvenance stays attached where available.

Source ecosystem

Source context stays visible.

Source matrix
Representative resources shown below.

Representative NosoGraph source integrations and their integration maturity

Source	Role in NosoGraph	Integration maturity
MONDO	Disease ontology and identifiers	STABLE
HPO / HPOA	Phenotypes and annotations	STABLE
PubMed	Literature and evidence workspace	STABLE
ClinicalTrials.gov	Clinical study records	STABLE
Open Targets	Target–disease data and connector	BETA
GWAS Catalog	Genetic associations	BETA
bioRxiv / medRxiv	Preprint evidence workspace	EXPERIMENTAL

Integration maturity describes NosoGraph's implementation state, not the scientific quality or clinical validity of an upstream resource. View the complete source matrix →

Two surfaces, one project

This site is the documentation. The dashboard is local software.

GitHub Pages hosts only these docs. The NosoGraph dashboard, CLI, and API run on your machine — every capture shown here comes from a local run, and there is no hosted demo to log into.

Evidence Explorer — a local dashboard surface for inspecting condition context, claims, evidence direction, and source-linked research data. LOCAL RUN · PUBLIC ALPHA

Researcher workflow

Start with a question. Follow the evidence.

A practical path from disease context to the next inspectable research surface.

1. 01
   Choose context

   Start with a disease module or research question, then check its curation and coverage context.

   SLE walkthrough →
2. 02
   Inspect claims

   Open the condition view and identify the exact relationship type instead of reading an unlabeled edge.

   Evidence Explorer guide →
3. 03
   Review direction

   Read SUPPORTS, CONTRADICTS, INCONCLUSIVE, or UNASSERTED as categorical evidence semantics.

   Evidence concepts →
4. 04
   Trace context

   Follow study/source metadata and the provenance chain; keep unavailable fields explicitly unknown.

   Evidence tracing →
5. 05
   Compare if useful

   Compare two to five conditions with explicit missingness when a cross-condition question is appropriate.

   Compare workflow →

Research software for investigation — not medical advice or clinical decision support. Research-use boundary →

Developer path

Run it locally. Inspect the model. Extend the system.

NosoGraph is source-install software: use the CLI, local FastAPI surface, structured disease modules, and documented interfaces to inspect or extend research workflows.

Run locallysource install

git clone https://github.com/AdamEddahmouni/nosograph.git
cd nosograph
cp .env.example .env
docker compose --profile full up --build

Then open the local dashboard at http://localhost:8000.

CLIValidate disease modules and run task-oriented analysis commands.CLI guide →

APIQuery structured disease, evidence, provenance, and comparison surfaces.API reference →

ValidationRun the local project gate before contributing.make ci-local

SourcesExtend source integrations with licensing, provenance, maturity, and tests.Source contributions →

DATADisease modulesstructured domain objects

MODELClaims · evidencetyped relationships + context

INTERFACESCLI · API · webinspect, validate, compare

EXTENDModules · adapters · testsdocumented contribution surfaces

Open source

Contribute to the parts you can inspect.

The project accepts focused improvements to code, curation, source context, documentation, and validation.

BuildImprove software and interfaces without weakening the tests.Code contribution guide →

CurateImprove disease modules with sourced evidence, provenance, and licensing discipline.Disease curation guide →

IntegrateAdd or maintain source entries with terms, maturity, and connector tests.Source contribution guide →

DocumentStrengthen research examples and accessible project documentation.Contributing overview →

Good first issues → are listed with their intended area; governance and decision-making are documented publicly.

Disease Intelligence. Connected.

The research model is open to review — read the source, check the current status, or cite the project.

Apache-2.0 source · upstream data terms vary by source · v0.2.1 · Concept DOI 10.5281/zenodo.22055279