{"architecture":{"family":"Evoformer + structure module (attention over MSA and pair representations)","input":"amino-acid sequence (+ multiple-sequence alignment, templates)","output":"3D coordinates, pLDDT, PAE","params":"~93M","pretraining":"supervised on PDB structures + self-distillation on predicted structures"},"article":null,"cancer_slugs":["pan-cancer"],"category":"protein-structure","confidence":"high","datasets":[{"name":"Protein Data Bank (wwPDB / RCSB)","note":"","role":"training","slug":"pdb"}],"developer":"Google DeepMind","evaluation":[{"benchmark":"CASP14","external":true,"metric":"GDT_TS","source":"https://www.nature.com/articles/s41586-021-03819-2","value":"median 92.4 across targets"}],"hf":null,"kind":"foundation","license":"Apache-2.0 (code) / CC-BY-4.0 (parameters)","limitations":"- Static single conformation; poor for intrinsically disordered regions and many point-mutation effects.\n- No ligands or complexes with nucleic acids in AlphaFold 2 (see AlphaFold 3).","links":{"demo":"https://alphafold.ebi.ac.uk/","docs":null,"doi":"10.1038/s41586-021-03819-2","github":"https://github.com/google-deepmind/alphafold","huggingface":null,"paper":"https://www.nature.com/articles/s41586-021-03819-2","pmid":null},"modalities":["protein-sequence"],"name":"AlphaFold 2","notable_uses":"CASP14 winner (2020); more than 30% of publications citing AlphaFold concern disease research, including analyses of mutated cancer proteins; feeds Isomorphic Labs' drug-design work.","openness":"open-weights","regulatory":{"intended_use_en":"Research tool.","intended_use_pl":"Narz\u0119dzie badawcze.","source_url":"https://github.com/google-deepmind/alphafold","status":"not-applicable"},"regulatory_status":"not-applicable","release_date":"2021-07-15","run_snippet":"# fastest path: ColabFold (AlphaFold2 + MMseqs2)\npip install colabfold[alphafold]\ncolabfold_batch input.fasta out_dir/\n# or query precomputed structures: https://alphafold.ebi.ac.uk/entry/<UniProt accession>","settings":["basic-research","drug-discovery"],"slug":"alphafold","sources":[{"label":"Jumper J et al. Highly accurate protein structure prediction with AlphaFold. Nature 2021","url":"https://www.nature.com/articles/s41586-021-03819-2"},{"label":"AlphaFold Protein Structure Database","url":"https://alphafold.ebi.ac.uk/"},{"label":"Google DeepMind \u2014 AlphaFold","url":"https://deepmind.google/science/alphafold/"}],"summary":"Predicts the 3D structure of a protein from its amino-acid sequence at near-experimental accuracy; the AlphaFold Protein Structure Database (with EMBL-EBI) provides predicted structures for over 200 million proteins, including cancer-relevant targets and mutants.","tasks":["protein-structure"],"training":{"institutions":"DeepMind; PDB, UniProt as data sources","summary_en":"Experimentally determined structures from the Protein Data Bank (cut-off 2018) plus self-distillation on predicted structures of UniRef sequences.","summary_pl":"Struktury wyznaczone eksperymentalnie z Protein Data Bank (do 2018) plus samodestylacja na przewidywanych strukturach sekwencji UniRef."},"updated_at":"2026-09-05T22:26:00.969983","url":"/ai-oncology/models/alphafold","usage":{"hardware":"GPU; MSA search is the slow part \u2014 ColabFold with MMseqs2 is the practical route","library":"JAX (repo) / ColabFold","snippet":"# fastest path: ColabFold (AlphaFold2 + MMseqs2)\npip install colabfold[alphafold]\ncolabfold_batch input.fasta out_dir/\n# or query precomputed structures: https://alphafold.ebi.ac.uk/entry/<UniProt accession>"},"verified_at":"2026-09-05T22:26:00.969670","verified_by":"editorial","version":"2","what_it_does":"Sequence in, atomic coordinates and per-residue confidence (pLDDT) out. In oncology it underpins structural interpretation of driver mutations, target druggability assessment and structure-based drug design."}