AlphaFold Solves Protein Folding
DeepMind's AI cracks a 50-year-old grand challenge in biology, predicting the 3D structure of every known protein.
Explore this event on the interactive timeline →In December 2020, DeepMind's AlphaFold 2 solved the protein folding problem—predicting the 3D structure of a protein from its amino acid sequence with experimental-level accuracy. In 2022, AlphaFold predicted the structure of virtually all 200 million known proteins and released the database for free. This is Kurzweil's GNR convergence in action: the "R" revolution (AI) directly accelerating the "G" revolution (Genetics/Biology).
Key Numbers
- Proteins Predicted
- 200+ Million
- CASP14 Accuracy (GDT)
- 92.4 (vs. 60 prior)
- Traditional Cost/Protein
- $100,000+
- AlphaFold Cost/Protein
- ~$0
Verified Facts
- Protein folding was considered a "50-year grand challenge" in biology. Experimental determination of a single protein structure could take years and cost $100,000+.
- AlphaFold 2 achieved a median GDT score of 92.4 at CASP14 (2020), compared to the ~60 score of the best previous methods. Scores above 90 are considered "competitive with experimental methods."
- By July 2022, AlphaFold had predicted the structure of 200+ million proteins—virtually every protein known to science—and released the database free of charge.
- AlphaFold 3 (2024) extended predictions to all biomolecular interactions: proteins with DNA, RNA, ligands, and small molecules. This enables AI-driven drug design.
- This validates Kurzweil's prediction: "Drug design by simulation—using AI to simulate molecular interactions and design drugs in silico" (The Singularity Is Near, Chapter 5).