OpenCure Labs — Download Research Data

Quick Links

📋 Latest 100 Results (JSON)

A rolling cache of the most recent published results — ideal for quick integration.

curl https://pub.opencurelabs.ai/latest.json | python -m json.tool

📊 Live Statistics

Real-time counts of total results, novel findings, and active skill types.

curl https://ingest.opencurelabs.ai/results/count

REST API

The ingest API at ingest.opencurelabs.ai supports paginated queries with filters. No API key required for read access.

GET /results

Query published results with optional filters and cursor-based pagination.

Parameter	Type	Description
`limit`	int	Results per page (default 50, max 200)
`skill`	string	Filter by skill type
`species`	string	Filter by species (`human`, `dog`, `cat`)
`novel`	bool	Filter novel findings only (`true`/`false`)
`date`	string	Filter by date (`YYYY-MM-DD`)
`after`	ISO timestamp	Cursor for pagination — pass `next_cursor` from previous response

# First page
curl "https://ingest.opencurelabs.ai/results?limit=100"

# Next page (use next_cursor from response)
curl "https://ingest.opencurelabs.ai/results?limit=100&after=2026-03-24T12:00:00.000Z"

# Novel neoantigen results only
curl "https://ingest.opencurelabs.ai/results?skill=neoantigen_prediction&novel=true&limit=50"

Response includes has_more (boolean) and next_cursor (ISO timestamp) for pagination.

GET /results/count

Get aggregate statistics without fetching full results.

curl "https://ingest.opencurelabs.ai/results/count"

# Response:
# { "total": 3300, "novel": 331, "today": 150, "skills": 7 }

GET /critiques

Query Grok's scientific reviews of results.

curl "https://ingest.opencurelabs.ai/critiques?limit=50"

# Filter by result
curl "https://ingest.opencurelabs.ai/critiques?result_id=RESULT_UUID"

Available Skills

Skill	Description
`neoantigen_prediction`	Tumor neoantigen binding prediction (MHC I/II)
`variant_pathogenicity`	Genetic variant pathogenicity scoring
`molecular_docking`	Protein-ligand docking simulations
`qsar`	Quantitative structure-activity relationship models
`structure_prediction`	Protein 3D structure prediction
`sequencing_qc`	Sequencing data quality control
`grok_research`	AI literature review and hypothesis generation

Full Result Objects (R2)

Each result's r2_url field points to the complete JSON object stored in Cloudflare R2, including full result_data and batch_critique from Grok's review.

# Get a result URL from the API, then fetch full data
curl "https://pub.opencurelabs.ai/results/neoantigen_prediction/2026-03-24/UUID.json"

Python Example

import requests

API = "https://ingest.opencurelabs.ai"

# Paginate through all results
all_results = []
cursor = None
while True:
    params = {"limit": 200}
    if cursor:
        params["after"] = cursor
    resp = requests.get(f"{API}/results", params=params).json()
    all_results.extend(resp["results"])
    if not resp["has_more"]:
        break
    cursor = resp["next_cursor"]

print(f"Downloaded {len(all_results)} results")

License

All research data is released under the MIT License. Results are AI-generated and should be independently validated before any clinical application.