AlphaFold
- (Harvard University - Harvard Taiwan Student Association)
- Overview
Announced on 8 May 2024, AlphaFold 3 was co-developed by Google DeepMind and Isomorphic Labs, both subsidiaries of Alphabet. AlphaFold 3 is not limited to single-chain proteins, as it can also predict the structures of protein complexes with DNA, RNA, post-translational modifications and selected ligands and ions.
AlphaFold 3 is an artificial intelligence (AI) tool that can predict the structure and interactions of proteins and other biological molecules. It was developed by Google DeepMind and Isomorphic Labs in 2024, and is considered to be more accurate than its predecessor, AlphaFold 2.
AlphaFold 3 can predict the 3D folded structure of proteins based on their amino acid sequences, and how they interact with other molecules like DNA, RNA, and small molecules called ligands. It can also predict the structures of complexes that proteins may form with other molecules. AlphaFold 3 uses a diffusion network, which is similar to the tools used in AI image generation, to generate accurate 3D models of biomolecular complexes.
AlphaFold 3 has a web server with a simple interface that doesn't require coding. Anyone with a Google account can use the server to input the name of a protein or nucleic acid and generate structure predictions. Google claims that AlphaFold 3 improves on existing prediction methods by at least 50% for protein interactions with other molecules, and doubles prediction accuracy for some important categories of interaction.
AlphaFold 3's capabilities could lead to transformative science, such as accelerating drug design and genomics research, developing biorenewable materials, and creating more resilient crops. For example, AlphaFold 3's prediction of a spike protein from a common cold virus interacting with antibodies and simple sugars accurately matches the true structure. This could help researchers better understand coronaviruses, including COVID-19, and potentially improve treatments.
The AlphaFold 3 program powers the AlphaFold server, which offers a web interface that allows non-coders to input the name of a protein or a nucleic acid and generate structure predictions, including critical information about joint structures with elements such as RNA, DNA, ligands and others.
AlphaFold can predict an array of biological interactions and structures. Unlike its predecessor, AlphaFold 2, which relied on an architecture optimized for predicting the structure of individual proteins, AlphaFold 3 (AF3) uses a diffusion model that predicts raw atom coordinates.
[More to come ...]
