AlphaFold 3 Advances Medical Research

Google recently released AlphaFold 3, an AI model that predicts the structure and interactions of molecules to help researchers better understand the biological world in the hopes of advancing medical research.

An article in Nature introduced AlphaFold 3 to the scientific world. Developed jointly by Google DeepMind and Isomorphic Labs, the paper discusses how this new model has “significantly improved accuracy over many previous specialized tools.” The new model has demonstrated an overall 50% improvement over other existing models in predicting methods. In some “important categories of interaction,” it has doubled in accuracy.

AlphaFold 3 is available for scientists, free of charge, through the newly released AlphaFold Server for non-commercial research. Scientists can leverage the computation resources of AlphaFold to help them generate hypotheses that they can test in their labs, which can significantly accelerate the creation of novel studies.

Isomorphic Labs is also partnering with pharmaceutical companies to use AlphaFold 3 to support current drug design projects by researching new ways to target disease or by adapting existing methods.

First released in 2020, AlphaFold 2 contributed to research in malaria vaccines and cancer treatments, focusing on understanding protein structures. Cited over 20,000 times, it won the Breakthrough Prize in Life Sciences in 2023.

The new model predicts additional molecules beyond proteins and has applications not only in drug creation, but also in the areas of biorenewable materials, the development of resilient crops, and research in genomics.

AlphaFold 3 can create complex 3D images of the various molecules, including proteins, DNA, RNA, and ligands. It can model the impact of chemicals and antibodies on the various components of cells. AlphaFold 3’s real breakthrough is its ability to “compute entire molecular complexes in a holistic way.”

Created in 2021, Isomorphic Labs is a subsidiary of Alphabet that aims to leverage AI and other computational methods in drug discovery. It works hand-in-hand with Google DeepMind to develop AlphaFold.