The University of Washington’s David Baker and his team have made a significant breakthrough in artificial intelligence-powered molecule design. Following their Nobel Prize win last year, they aim to develop new enzymes from scratch that can catalyze nearly any scientific experiment.
Enzymes are proteins that speed up chemical reactions under mild conditions, outperforming even the best human chemists. Baker’s team has been working on creating novel enzymes for decades using traditional methods of modifying naturally occurring proteins and nudging them toward their molecular goals over time.
However, these approaches have limitations. According to Sam Pellock, an acting instructor in the Baker Lab, traditional enzyme design is like buying a suit from a thrift store – the fit will likely be off. With AI, researchers can now tailor-make enzymes to ensure a perfect fit for every step of the reaction.
As an initial demonstration, the team used their RFdiffusion AI model to generate enzymes designed to break ester bonds, which are crucial in forming the underlying structure of many common substances. The success of these enzymes was measured by tracking the cleaving of bonds through identifiable signals using fluorescence and other means.
The researchers were able to develop enzymes that employed a previously known mechanism of action but used molecular structures not found in nature. In their experiments, they found that these AI-designed enzymes can be quite efficient, with room for improvement due to the complexity of chemical transformations involved.
Source: https://www.fiercebiotech.com/medtech/after-ai-antibodies-and-peptides-uws-baker-lab-now-tackles-computer-generated-enzymes