Member
- Joined
- Oct 14, 2023
- Messages
- 225
- Thread Author
- #1
How artificial intelligence helps create the medicine of the future.
Almost five years ago, Google's DeepMind research lab introduced AlphaFold — an artificial intelligence system capable of predicting protein structures in the human body. In 2020, an improved version was released — AlphaFold 2.
DeepMind today announced the latest release of AlphaFold, which can generate predictions for almost all molecules in the Protein Data Bank — the world's largest open database of biological molecules. In addition, Isomorphic Labs, a subsidiary of DeepMind, is already using the new AlphaFold model for drug development.
New features
AlphaFold is no longer limited to predicting protein structures. DeepMind claims that the model can also accurately predict the structures of ligands-molecules that bind to "receptor" proteins, as well as nucleic acids and post-translational modifications. These capabilities can be a valuable tool in the field of drug development, as they will help scientists identify and design new molecules that can become the basis for new drugs.
Previously, researchers used computer simulations known as "docking techniques" to determine the interaction of proteins and ligands. However, with the latest version of AlphaFold, there is no need to use this method. DeepMind claims that the level of modeling that AlphaFold provides is unattainable for modern docking methods.
The DeepMind message says: "Our early analysis shows that our model significantly outperforms the previous version of AlphaFold in solving protein structure prediction problems relevant to drug development, such as antibody binding."
However, the latest AlphaFold is not without its drawbacks. Researchers from DeepMind and Isomorphic Labs point out that the system does not always cope with predicting the structures of RNA molecules. Both companies are actively working to fix this issue.
Almost five years ago, Google's DeepMind research lab introduced AlphaFold — an artificial intelligence system capable of predicting protein structures in the human body. In 2020, an improved version was released — AlphaFold 2.
DeepMind today announced the latest release of AlphaFold, which can generate predictions for almost all molecules in the Protein Data Bank — the world's largest open database of biological molecules. In addition, Isomorphic Labs, a subsidiary of DeepMind, is already using the new AlphaFold model for drug development.
New features
AlphaFold is no longer limited to predicting protein structures. DeepMind claims that the model can also accurately predict the structures of ligands-molecules that bind to "receptor" proteins, as well as nucleic acids and post-translational modifications. These capabilities can be a valuable tool in the field of drug development, as they will help scientists identify and design new molecules that can become the basis for new drugs.
Previously, researchers used computer simulations known as "docking techniques" to determine the interaction of proteins and ligands. However, with the latest version of AlphaFold, there is no need to use this method. DeepMind claims that the level of modeling that AlphaFold provides is unattainable for modern docking methods.
The DeepMind message says: "Our early analysis shows that our model significantly outperforms the previous version of AlphaFold in solving protein structure prediction problems relevant to drug development, such as antibody binding."
However, the latest AlphaFold is not without its drawbacks. Researchers from DeepMind and Isomorphic Labs point out that the system does not always cope with predicting the structures of RNA molecules. Both companies are actively working to fix this issue.