Algorithmic Alchemy: Transforming Drug Discovery with Open AI
summary
Argues that AlphaFold-style models truly change preclinical work by predicting 3D protein structures with near-experimental accuracy — GDT score 92.4 — but still miss chemistry that matters: ligands, multimer context and dynamics. Notes community reimplements AlphaFold in PyTorch (OpenFold), adds ligand transplant tools (AlphaFill), and demonstrates a hemoglobin prediction that places heme groups correctly. Flags commercialization limits: closed code, UI quota 20 predictions/day, and industrial deals paying tens of millions (about €38–€45M upfront).
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:39AlphaFold protein folding
- 10:48AlphaFold limitations open source
- 18:12AI drug discovery hype
- 20:19Pharma tech convergence
- 23:12Open source sustainability
- 26:45Pharma drug development
- 29:13OpenFold role consortium
- 32:47Motivation future challenges
- 34:59Conclusion
related talks
29mProtein folding and what it means for drug discovery
04mHow Machine Learning tools at Google made AlphaFold possible
53mHow machine learning is influencing protein engineering
01h 13mDeep Learning in Structural Biology and Protein Design: How, Where, and Why
14mAI Is Built to Be Applied
06mTools vs Products: Lessons learned from a niche industry
41mThe kaleidoscopic futures of AI and security
16mBeyond Patterns: The Case for Physics-Grounded AI in Drug Discovery