Google DeepMind marks five years of AlphaFold's impact on biology

Marking five years since AlphaFold 2 solved protein structure prediction at the CASP14 competition in 2020, Google DeepMind published a retrospective on the tool’s use across biology and medicine. The company said the AlphaFold Protein Structure Database — launched in 2021 with EMBL-EBI and expanded a year later to cover more than 200 million predicted structures, close to every protein catalogued by science — had been used by researchers in more than 190 countries, including a substantial number in low- and middle-income countries, and cited directly or through its methodology in over 35,000 published papers.

DeepMind highlighted disease research as a major application, saying more than 30% of AlphaFold-related research addressed understanding disease, and that AlphaFold-based research was about twice as likely to be cited in clinical literature as typical structural-biology work. It pointed to AlphaFold 3, which extended prediction to molecular interactions relevant to drug design, and to Isomorphic Labs, DeepMind’s drug-discovery spinout, as evidence the underlying method was moving from academic tool toward pharmaceutical application. The achievement had already been recognised outside the company: Demis Hassabis and John Jumper shared the 2024 Nobel Prize in Chemistry for the original work.

As a retrospective rather than a new result, the post is a marker of AlphaFold’s sustained role as one of AI’s most widely cited applications outside language and image models — the case most often invoked, five years on, as evidence that deep learning had produced a durable scientific tool rather than a one-off benchmark result.