NosoGraph Public-Readiness Baseline¶

Audit date: 2026-08-20
Starting HEAD: 063966540ea217bab8e4ff489830433fced2e3c9
Branch: master (clean working tree, tracking origin/master)
Remote: https://github.com/AdamEddahmouni/med-research.git
Transformation target: NosoGraph — The Open Computational Map of Human Disease

Repository state at baseline¶

Item	Value
Package name (PyPI/CLI)	`med-research`
Python import path	`med_research`
Version	`2.1.0`
License (baseline)	MIT
Python	3.11–3.12
Disease module directories	10,407
Default branch	`master`
CI workflow	`.github/workflows/test.yml` (lint, security, test 3.11/3.12, integration)

Technology stack¶

Runtime: Python 3.11+, FastAPI, Uvicorn, Celery, Redis
Data: SQLite (evidence workspace, universal biomedical store), DuckDB analytics, JSON disease KGs
Frontend: Vanilla JavaScript dashboard (src/med_research/web/static/)
Testing: pytest, pytest-xdist, Playwright (slow tier), ruff, mypy (informational)
Lock files: requirements-lock.txt, requirements-dev-lock.txt (68 packages verified by scripts/lock_verify.py)
Containers: Dockerfile, docker-compose.yml

Repository layout (high level)¶

src/med_research/          # Application source (CLI, diseases, pipeline, biomed, web)
tests/                     # Unit, integration, browser tests
docs/                      # Usage docs + superpowers design archives
scripts/                   # Lock verify, disease batch, biomed imports
data/                      # Runtime DBs (gitignored); reports under data/reports/
.github/                   # CI, issue/PR templates, CODEOWNERS

Disease corpus maturity (honest baseline)¶

Tier	Count (approx.)	Meaning
L0	Majority of 10k+	Open Targets scaffold; incomplete KG/config
L1	Subset	Partial KG or config gaps
L2	~45 promoted + 18 legacy curated	Strict validation pass (symptoms, queries, CAR-T, safety)
L3	23 (`CURATED_CONSENSUS_DISEASES`)	L2 + hand-curated GEO expression consensus
CI gate	8 only	`sle`, `ra`, `ibd`, `ms`, `ss`, `ssc`, `t1d`, `ad`

Important: README baseline overclaimed “10,403+ disease modules” as research-ready. Scaffold count ≠ curated count.

Biomedical model (baseline)¶

Per-disease JSON KGs: profile, genes, drugs, pathways, relationships
Disease configs in config.py (symptoms, PubMed/trial/GWAS queries, CAR-T, safety, screening)
Universal Biomedical Schema v1 (med_research.biomed): MONDO, HPO, HPOA, GO, Reactome, Uberon, ClinVar, openFDA imports
Evidence Workspace: multi-source adapters, deterministic claims, optional LLM enrichment
Provenance: pipeline/provenance.py (schema v1.0, SHA-256 fingerprints)

Evidence & provenance (baseline)¶

STABLE: Provenance metadata, coverage contracts, evidence deduplication
BETA: Evidence Workspace dossier ranking, LLM extraction (optional, requires API key)
EXPERIMENTAL: Some pipeline modules default to sle disease context in CLI defaults

Security baseline¶

.env gitignored; .env.example documents secrets (API_KEY, AUTH_SESSION_SECRET, optional OPENAI_API_KEY)
Secret pattern grep: no hardcoded API keys or AWS tokens found in tracked files
Bandit + pip-audit in CI
Research-only framing in SECURITY.md; no PHI by design

Licensing baseline¶

Source: MIT (to migrate to Apache-2.0)
Third-party data: documented in docs/licensing.md (MONDO CC BY 4.0, HPO custom, Open Targets, NCBI, etc.)
Disease JSON compilations: project license for schema/compilation; underlying facts subject to source terms

Documentation baseline¶

Present	Missing (pre-transformation)
README, CONTRIBUTING, SECURITY, CODE_OF_CONDUCT, CHANGELOG	GOVERNANCE, ROADMAP, CITATION.cff
docs/licensing.md, deployment.md, api-reference.md	docs/audits/, docs/architecture/, docs/legal/*
docs/public-launch.md, disease-curation.md	data/sources/registry
RELEASING.md	NOTICE (Apache attribution)

CI & quality baseline¶

make ci-local: ruff check/format, lock verify, import audit, serial offline pytest (80% coverage gate in CI)
make test-offline: unit tests, no network
Disease validate --all --strict: expected non-zero on 10k scaffold registry (documented in AGENTS.md)
mypy: informational only (TECHNICAL_DEBT_ISSUES.md backlog)

Branding baseline¶

Surface	Current name	Target
Public docs / README	Medical Research Platform / med-research	NosoGraph
Python package	`med_research`	KEEP_FOR_COMPATIBILITY (document alias)
CLI entry point	`med-research`	KEEP_FOR_COMPATIBILITY
GitHub remote	`med-research`	Rename deferred (not in scope)

Secrets & data governance scan (baseline)¶

No .env tracked
Runtime DB paths gitignored (data/*.sqlite3)
No PHI patterns audited in repo content
OpenAI key placeholders commented in .env.example only

Validation commands to run post-transformation¶

make venv-sync          # or pip install -r requirements-lock.txt && pip install -e .
make ci-local
python -m med_research.cli disease validate sle --strict
python -m med_research.cli disease validate ra --strict
# ... remaining curated eight

This document is the Wave 0 snapshot. See companion audits under docs/audits/ for gap analysis and remediation tracking.