Five years after its release, Google’s protein-mapping AlphaFold2 large-language model has “revolutionized science,” Nature reported, helping to boost key biological research from drug discovery to cancer analysis.
Proteins are long molecules made of sequences of smaller ones. Their vital role in the body is dictated by their 3D shape, and for decades scientists have tried to predict how they “fold” from their 2D sequence. AlphaFold2 solved that problem in 2020, and scientists can now predict almost any protein’s shape with incredible precision.
More than 3 million people in 190 countries have used the database, which is cited in thousands of studies. “We use it for every project,” one biologist, who deployed AlphaFold to understand how sperm binds to eggs, said.