A groundbreaking study conducted by researchers from IBM, Oxford University, and Diamond Light Source demonstrates the potential of IBM’s AI model, MoLFormer, in generating antiviral molecules. These molecules have shown efficacy against multiple target virus proteins, including SARS-CoV-2, the virus responsible for COVID-19. The findings, published in Science Advances, highlight how AI can accelerate the drug discovery process, paving the way for faster response to future pandemics.

The collaboration between IBM, Oxford University, and Diamond Light Source began when a group of computer scientists at IBM set out to explore the possibilities of generative AI in designing novel molecules to combat SARS-CoV-2. Initially skeptical, experts like David Stuart from Oxford University, renowned for his work on HIV, SARS, and Ebola, joined forces with IBM and embarked on a three-year journey. Their goal was to demonstrate that generative AI could identify viable starting points for antivirals through collaboration with Enamine Ltd., a chemical supplier, and other researchers at Oxford.

What sets MoLFormer apart is its foundation model, which was pre-trained on vast amounts of raw data, allowing it to create new inhibitors for multiple protein targets without additional training or knowledge of the targets’ 3D structures. The team focused on two essential SARS-CoV-2 proteins, the spike protein and the main protease. Utilizing these targets, they swiftly identified four potential COVID-19 antivirals, significantly reducing the time typically required with conventional methods. Diamond Light Source’s high-throughput macromolecular crystallography beamlines were instrumental in visualizing how a subset of AI-generated compounds bound to the main protease.

While the validated molecules must undergo further hurdles, including clinical trials, before they can become drugs, this breakthrough showcases the crucial role of generative AI in future drug development. The study’s co-senior author, Payel Das from IBM Research, emphasized the potential of generative AI in expediting the search for new treatments when faced with future viruses.

The successful deployment of MoLFormer reinforces the importance of generating compounds that bind effectively to drug targets, thereby accelerating the drug discovery pipeline. Martin Walsh, co-senior author at Diamond Light Source, expressed the significance of this breakthrough in enhancing our preparedness for future pandemics.

The researchers’ efforts highlight the importance of being prepared for emerging viruses by developing drugs that act on different sites of the protein. This approach makes it significantly harder for the virus to evade treatment options. The study’s findings hold promise not only for COVID-19 but also for combating other mutating viruses like the flu or even yet-to-emerge viruses.

IBM’s AI model represents a significant leap forward in the fight against COVID-19 and offers hope for expedited drug development in times of crisis. With the ability to generate starting points for antivirals efficiently, the AI model showcases its potential to address urgent, life-threatening illnesses and fortify our defenses against future pandemics.

By Impact Lab