Researchers at Google DeepMind have announced a major breakthrough in predicting how drug molecules interact with protein targets in the human body. The new AI system, called AlphaFold-Drug, can accurately predict binding affinity and potential side effects of candidate drug molecules with 89 percent accuracy.
The system builds on the AlphaFold protein structure prediction platform and adds the ability to simulate molecular dynamics at speeds thousands of times faster than traditional computational chemistry methods. This could reduce the drug discovery timeline from years to months.
Three pharmaceutical companies, including Roche and Novartis, have already signed licensing agreements to integrate AlphaFold-Drug into their research pipelines. DeepMind plans to make a limited version available to academic researchers for free.
