Pharma consortium trains OpenFold3 on 20,000 proprietary structures, beats Boltz-2 with 50% accuracy

TheMoonMidas · x · 2026-09-14

A Nature news report describes a pharma consortium (AbbVie, Astex, and others) fine-tuning OpenFold3, an open-source replication of AlphaFold 3, on more than 20,000 proprietary protein structures from their internal databases.

• The resulting model achieved high-accuracy predictions on 50% of test cases, vs 40% for Boltz-2 and 33% for OpenFold.
• The edge comes from private data: protein–drug interaction structures locked in company vaults that public datasets lack.
• Findings show pharma's secret data markedly outperforms models trained on public data alone or on individual siloed datasets.

Related event: Five Pharma Firms Federally Fine-Tune OpenFold3 on Private Structures, Beating Public-Data Models(6 posts)→

Original post →

More from Research

• Spatial Hub aggregates ~5000 public COLMAP-compatible spatial datasets for world models — janusch_patas · 2026-10-31
• ARC Prize 2026 closes in 3 days as Yi-Chia Chen tops ARC-AGI-3 at 75.56% — GregKamradt · 2026-10-31
• NVIDIA Nemotron 3.5 scales LLM self-verification, +9-10 points on agentic benchmarks — sarahcat21 · 2026-10-31
• Unofficial OpenTargets MCP Server Brings Gene-Drug-Disease Data to LLMs — modelcontextprotocol · 2026-10-31
• Researchers find 5 personality structures in LLM weight space, Big Five-like — luke_drago_ · 2026-10-31
• NVS benchmarks are misaligned: proper pose alignment puts 3D Gaussians back at SOTA — kwangmoo_yi · 2026-10-31

Research channel →