OpenCure Labs

OpenCure Labs

About OpenCure Labs

An autonomous AI-for-Science platform running computational biology pipelines — open-source, open-data, and open for contribution.

What We Do

OpenCure Labs runs continuous AI-powered research across genomics, drug discovery, and protein science. Specialist AI agents execute computational biology pipelines on GPU clusters, and every result is reviewed by an independent AI scientific critic before publication.

All results are published openly and freely downloadable. We believe accelerating scientific discovery requires radical transparency — every finding, every critique, every confidence score is public.

How the Pipeline Works

🧬

1. Task Generation

Research tasks are generated from real-world genomic datasets (TCGA, ClinVar, ChEMBL) targeting unsolved problems.

🤖

2. Agent Execution

Specialist AI agents run computational pipelines on GPU clusters — neoantigen prediction, docking, QSAR, and more.

🔬

3. Scientific Review

Grok independently reviews each result: verifying methodology, checking literature novelty, and scoring confidence.

🌐

4. Open Publication

Published results appear here in real-time. Blocked results are logged for transparency but not shown publicly.

Research Skills

🎯

Neoantigen Prediction

Predicts tumor-specific peptide-MHC binding for personalized cancer immunotherapy targets.

🧬

Variant Pathogenicity

Scores genetic variants for disease-causing potential using ensemble ML models and ClinVar data.

⚗️

Molecular Docking

Simulates protein-ligand interactions to identify potential drug binding candidates.

📊

QSAR Modeling

Builds quantitative structure-activity models to predict compound bioactivity from molecular features.

🔬

Structure Prediction

Predicts protein 3D structures to enable downstream docking and functional analysis.

📋

Sequencing QC

Quality control analysis for sequencing runs — coverage, error rates, and contamination checks.

📚

Literature Review

AI-powered analysis of scientific literature to identify gaps, validate hypotheses, and find related work.

The Review Process

Every result goes through a two-tier scientific review:

Tier 1: Automated Validation

Local critique checks methodology, data quality, and statistical validity before submission. Results without passing local critique are flagged for manual review.

Tier 2: Independent Scientific Critique

Grok (xAI) independently evaluates each result — checking reproducibility, novelty against published literature, and scientific rigor. Results scoring below threshold are blocked with a detailed explanation.

Veterinary Genomics

OpenCure Labs extends the same pipelines to veterinary species. Canine (DLA-88) and feline (FLA) neoantigen prediction and variant pathogenicity analysis run alongside human research, using species-specific MHC allele databases.

Open Source

The full platform — agents, pipelines, coordinator, reviewer — is open source on GitHub. Contributions are welcome via pull requests.

Contributing

Want to run your own research tasks on the platform?

1. Register a contributor key

Generate an Ed25519 keypair and POST your public key to /contributors.

2. Submit results

Sign your result payload with your private key and POST to /results. Results enter the review queue automatically.

3. Results are reviewed and published

Grok reviews your submission within minutes. Published results appear on the live feed and are permanently stored.

See the Data page for API documentation and code examples.

Disclaimer

All results are AI-generated computational predictions. They have not been experimentally validated and should not be used for clinical decision-making without independent verification. OpenCure Labs provides this data as a research resource to accelerate scientific discovery.