Paving the Way to Precision Drugs

An AI-powered tool that could help scientists create better, more targeted medicine for people with cancer, Alzheimer’s and many other diseases is under development at the University at Buffalo.

The key, said UB researcher Thomas Grant—who received a $2.18 million federal grant to develop the tool—is understanding how proteins behave in the body.

The tool, SWAXSFold, is built around an advanced X-ray technique called SWAXS. Using SWAXS, scientists can extract information about proteins dissolved in water at room temperature, which is how they exist in human cells. This is important, said Grant, an assistant professor of structural biology in the Jacobs School of Medicine and Biomedical Sciences, because “proteins are constantly moving and changing shape. Current methods often give us static snapshots, but we need to see them in action to really understand how they work and how to design drugs that target them effectively.”

SWAXSFold aims to provide more precise information by integrating SWAXS data into an AI model inspired by AlphaFold, a Nobel Prize-winning program that predicts protein structures from amino acid sequences. The downside of AlphaFold is that it can’t determine which shapes proteins actually take under specific conditions. SWAXSFold closes that gap by anchoring predictions in experimental data.

According to Grant, no one has done this before—partly because they didn’t have the computing power of Empire AI, New York State’s $500 million research consortium advancing AI for the public good. Grant’s team, including postdoctoral associate Sarah Chamberlain and PhD student Patrick Oduro, is using Empire AI’s supercomputer, located at UB, to train and validate SWAXSFold. “We wouldn’t be able to do this level of computation without it,” he said.

The research is important for drug discovery because drugs need to bind to the actual shapes proteins take in a person’s body. This is especially important for proteins that are hard to develop drugs against, such as cancer-causing proteins that constantly morph.

Even more promising, Grant said new tools are being developed to show how disease-causing mutations reshape proteins. “If we can see exactly how a mutation alters a protein’s shape and function,” he said, “we can design therapies targeted to that specific change”—bringing truly personalized medicine one step closer to reality.