Developer API

A read-oriented REST API over the Curebiology dataset — research papers, news, blog posts, clinical trials, FDA approvals and nonprofits — plus AI question answering and MAMMAL-style biomedical predictions.

Overview

Base URL: your deployment origin (e.g. http://localhost:3000).
All data endpoints accept an optional disease slug (defaults to your selected disease).
Responses are JSON. List endpoints return { total, page, limit, items }.
Authentication is not required for public data endpoints.

Data endpoints

GET/api/stats?disease={slug}

Aggregated counts and distributions for a disease.

GET/api/papers?disease={slug}&q=&subtype=&page=

Paginated research papers.

GET/api/papers/{id}

A single research paper.

GET/api/news?disease={slug}&q=&page=

Paginated news items.

GET/api/blog?disease={slug}&q=&page=

Paginated blog posts.

GET/api/blog/{slug}

A single blog post.

GET/api/trials?disease={slug}&q=&location=

Clinical trials.

GET/api/fda?disease={slug}

FDA approvals for the disease.

GET/api/nonprofits?disease={slug}

Related nonprofit organizations.

GET/api/graph?disease={slug}&topic=&size=

RAG knowledge-graph nodes and clusters.

GET/api/ncbi?db=pubmed&term=&retmax=

Live proxy to NCBI E-utilities (PubMed/PMC).

Diligence engine

POST/api/diligence/analyze

Run a full multi-source diligence pass on a compound / disease / modality / target. Returns the complete report: signals, scores, analogs, failure modes, contradictions, risks.

{ "compound": "osimertinib", "disease": "non-small cell lung cancer", "target": "EGFR" }

POST/api/diligence/deck

Multipart upload of a pitch deck (PDF/TXT) + mechanism/target/molecule/trialData fields. Extracts claims, grades each against open sources, and returns the deck report.

GET/api/diligence/reports

List the 25 most recent diligence runs (metadata only).

GET/api/diligence/reports/{id}

Fetch a stored report by id (full payload).

Sources: PubMed (NCBI E-utilities), ClinicalTrials.gov API v2, openFDA (FAERS, Drugs@FDA), PubChem PUG-REST, Open Targets Platform GraphQL, ChEMBL Web Services, PatentsView. All free, key-optional endpoints; each analysis is fail-soft per source and every synthesized claim carries a source chain. Reports are heuristic — not investment or medical advice.

AI & model endpoints

POST/api/ai/ask

Ask a plain-English question; grounded in the indexed dataset. Uses OpenAI if configured, else a local fallback.

{ "question": "What are the latest immunotherapy advances?", "disease": "cancer" }

POST/api/mammal

MAMMAL-style predictions (illustrative). task = protein-protein | drug-target | clintox.

{ "task": "protein-protein", "proteinA": "EGFR", "proteinB": "GRB2" }

MAMMAL predictions are illustrative approximations for demo purposes and must not be used for clinical decisions. NCBI data is subject to NCBI's usage policies.