{"entity":{"id":"paper-alphafold2-jumper-nature-2021","kind":"paper","name":"AlphaFold 2: predicting protein structures to near-experimental accuracy","aka":[],"tldr":"DeepMind's neural network predicted protein structures at CASP14 with a median backbone error of under 1 angstrom, comparable to experimental methods, and the team released predicted structures for essentially every human protein within a year.","summary":"AlphaFold 2 combines multiple-sequence alignments, an attention-based Evoformer network and an equivariant structure module trained end to end on the Protein Data Bank. At the blind CASP14 assessment in 2020 it achieved a median GDT of about 92 and a median backbone RMSD95 of 0.96 angstrom on the hardest targets, versus 2.8 angstrom for the next-best method.\n\nThe model provides per-residue confidence estimates (pLDDT), enabling users to distinguish reliable regions from disordered or uncertain ones. The accompanying AlphaFold Protein Structure Database, built with EMBL-EBI, released predicted structures for the human proteome and later for more than 200 million proteins.\n\nFor cancer drug discovery, AlphaFold accelerated structure-based design for targets without crystal structures and, with AlphaFold 3 (2024) extending to protein-ligand and protein-nucleic acid complexes, is now a routine part of the target-to-lead pipeline.","asOf":"2026-09-08","links":[{"label":"DOI","url":"https://doi.org/10.1038/s41586-021-03819-2"},{"label":"AlphaFold Protein Structure Database","url":"https://alphafold.ebi.ac.uk"}],"tags":[],"related":["alphafold3","de-novo-protein-design","idea-multimodal-foundation-model"],"cancers":[],"sections":["drug-discovery","ai-computation"],"technologies":["ai-drug-design","structural-biology-infrastructure"],"targets":[],"drugs":[],"companies":[],"institutions":[],"pathways":[],"terms":[],"trials":[],"people":[],"bottlenecks":["b-undruggable-targets","b-translational-valley","b-ai-validation"],"keyPapers":[],"journals":["nature"],"dependsOn":[],"notes":[],"journal":"Nature","year":2021,"doi":"10.1038/s41586-021-03819-2","pmid":"34265844","authors":"Jumper J, Evans R, Pritzel A, et al.","paperType":"methods","findings":["CASP14: median backbone RMSD95 of 0.96 angstrom (95% CI 0.85-1.16) vs 2.8 angstrom for the next-best method","Median GDT score around 92 across CASP14 targets, the first time a computational method reached experimental-grade accuracy","Per-residue confidence (pLDDT) reliably flags disordered and low-confidence regions","Predicted structures for the entire human proteome released in 2021; over 200 million proteins by 2022"],"whatItMeans":"The shape of nearly every protein is now available to any researcher in seconds instead of years, which shortens the path from a cancer target to a designed molecule. It does not by itself produce drugs: binding pockets, dynamics and cellular context still need experiment.","caveats":["Predicts single static conformations; many drug targets (kinases, GPCRs, KRAS) move between states","Accuracy is lower for proteins without evolutionary homologues, disordered regions and multi-protein complexes","Does not predict effects of point mutations or ligand binding (AlphaFold 3 and other tools partly address this)","The 2021 model was released with a non-commercial licence for weights, later loosened"],"changedPractice":false},"route":"/key-papers/paper-alphafold2-jumper-nature-2021/","neighbours":{"technology":[{"id":"ai-drug-design","kind":"technology","name":"AI-driven drug & target discovery","route":"/technologies/ai-drug-design/"},{"id":"alphafold3","kind":"technology","name":"AlphaFold 3","route":"/technologies/alphafold3/"},{"id":"de-novo-protein-design","kind":"technology","name":"De novo designed protein binders","route":"/technologies/de-novo-protein-design/"},{"id":"structural-biology-infrastructure","kind":"technology","name":"Structural biology infrastructure (cryo-EM, synchrotrons, AlphaFold)","route":"/technologies/structural-biology-infrastructure/"}],"idea":[{"id":"idea-multimodal-foundation-model","kind":"idea","name":"Patient-level multimodal foundation models for treatment selection","route":"/ideas/idea-multimodal-foundation-model/"}],"section":[{"id":"ai-computation","kind":"section","name":"AI & Computation","route":"/fronts/ai-computation/"},{"id":"drug-discovery","kind":"section","name":"Drug Discovery Platforms","route":"/fronts/drug-discovery/"}],"bottleneck":[{"id":"b-ai-validation","kind":"bottleneck","name":"AI that is built but not validated or deployed","route":"/bottlenecks/b-ai-validation/"},{"id":"b-undruggable-targets","kind":"bottleneck","name":"The undruggable drivers","route":"/bottlenecks/b-undruggable-targets/"},{"id":"b-translational-valley","kind":"bottleneck","name":"The valley of death between lab and product","route":"/bottlenecks/b-translational-valley/"}],"journal":[{"id":"nature","kind":"journal","name":"Nature","route":"/journals/nature/"}],"person":[{"id":"faisal-mahmood","kind":"person","name":"Faisal Mahmood","route":"/people/faisal-mahmood/"}]}}