AlphaFold, five years on: a protein-structure tool now used by over 3 million researchers

Five years after the AlphaFold 2 system solved a problem that had stumped biologists for half a century, Google DeepMind published an anniversary review describing how the tool has spread through the global research community. According to the company, AlphaFold is now used by more than 3 million researchers across over 190 countries, including more than 1 million users in low- and middle-income countries.

The underlying science remains significant. In 2020, AlphaFold 2 won the CASP 14 protein-structure prediction contest, effectively cracking a 50-year-old challenge of predicting a protein's three-dimensional shape from its amino-acid sequence. The AlphaFold Protein Structure Database, built with EMBL-EBI, has since released predictions for more than 200 million protein structures. The work was recognized with the 2024 Nobel Prize in Chemistry.

DeepMind reports that AlphaFold has been cited in more than 35,000 papers, with over 30 percent of related studies focused on disease. The company highlights concrete applications including honeybee conservation, heart-disease research involving the apoB100 protein, and crop-resilience studies, illustrating reach across medicine, agriculture and conservation.

Caveats matter. AlphaFold predicts static structures with confidence scores, but predictions are not experimental proof, and the model is less reliable for some disordered regions, complexes and molecular dynamics. AlphaFold 3, released in 2024, extends predictions to DNA, RNA and ligands but its code is available to academics only, not for commercial use. Researchers stress that predictions still require laboratory validation. Even with those limits, the five-year record shows a clear pattern: an AI tool that accelerates routine structural work, freeing scientists to focus on harder questions.