Accurate protein structure prediction across the known proteome (AlphaFold2)
Accurate protein structure prediction across the known proteome (AlphaFold2) is graded independently checked on whataifound.org, with the AI's role graded ai-led.
AlphaFold2 predicted three-dimensional structures for nearly all catalogued proteins from amino-acid sequence at accuracy rivalling experiment, work that earned the 2024 Nobel Prize in Chemistry.
- Verification
- Independently checked
- Autonomy
- AI-led
- Lab
- Google DeepMind
- Model
- AlphaFold2
- Field
- Biology
- Date
- 2021-07-15
- Human collaborators
- John Jumper, Demis Hassabis
- Problem posed
- 1972 · open 49 yrs
- Notability
- 19 Wikipedia language editions
What was found
AlphaFold2 pairs an attention-based neural network with evolutionary sequence information to predict how a protein folds from its amino-acid sequence. At the 2020 CASP14 blind assessment it reached accuracy competitive with experimental methods, and DeepMind and EMBL-EBI then released predicted structures for over 200 million proteins, nearly the entire catalogued proteome. Demis Hassabis and John Jumper shared the 2024 Nobel Prize in Chemistry for the work, alongside David Baker for computational protein design.
Novelty check
Protein-structure prediction had been an open grand challenge for roughly fifty years, benchmarked every two years by the CASP assessment. AlphaFold2's CASP14 result was a discontinuous jump over all prior methods and is uncontested.
Caveats
AlphaFold predicts structures rather than determining them experimentally; outputs are computational hypotheses that can be wrong for disordered regions, alternative folds, point mutations, and many complexes. It predicts natural structures, not new biology on its own. The grade reflects the prediction step; the Nobel recognised the human-built method.
Independent checks
CASP14 blind assessment (independent assessors): accuracy competitive with experiment across most targets
Royal Swedish Academy of Sciences (2024 Nobel in Chemistry): awarded to Hassabis and Jumper for protein structure prediction · link ↗
Sources
How this is graded
whataifound.org grades every entry on two axes: verification (how solid the result is, from a machine-checked proof down to refuted) and autonomy (how much the AI did versus its human collaborators). This finding is independently checked and ai-led. Full definitions are in the methodology.
Cite this entry
whataifound.org (2021). Accurate protein structure prediction across the known proteome (AlphaFold2). whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2021-07-15-alphafold2