Scientists at the University of Virginia School of Medicine have developed a new drug design method that uses AI diffusion models to identify and adapt to protein movement.
A new suite of artificial intelligence-powered tools called YuelDesign, YuelPocket, and YuelBond was created by a team at the University of Virginia School of Medicine to make the drug development process more efficient and identify new ways to make existing drugs more effective.
The average cost of developing a new drug is estimated to drop from hundreds of millions of dollars to more than $2.6 billion, and continues to rise. Approximately 90% of new drugs fail human testing.
This can be due to several factors, but one of the most important is the unpredictable reaction between drug molecules and their targets in the body, which can result in negligible or even negative effects of treatment.
Designing drugs that can respond to changes in protein shape
Proteins in the body are notorious for “induced seizures,” where they change shape after binding to drugs.
This phenomenon has traditionally been difficult to predict with computerized models, but the advanced AI “diffusion model” designed in this project can generate protein pocket structures and small molecules that fit into them, allowing for adaptive design.
“Think of it this way: Other methods try to design a key for a lock that is perfectly still, but inside the human body that lock is constantly shaking and changing shape. Because our AI designs the key while the lock is moving, the fit is more realistic,” said Dr. Nikolai V. Doholyan, of UVA’s Department of Neurology. “This could be a game-changer for patients suffering from cancer, neurological disorders, and many other conditions. Better drugs that target these unstable proteins are desperately needed, yet we continue to stall.”
The three tools work together to design the optimal shape.
YuelDesign uses diffusion models to design customized drug molecule shapes YuelPocket uses graph neural networks to pinpoint where drug molecules bind to proteins YuelBond ensures that chemical bonds in designed molecules are accurate
“Most existing AI tools treat proteins as frozen statues, but that’s not the case in biology. Our approach allows proteins and drug candidates to evolve together during the design process, just as they do in the body,” said researcher Dr. Jian Wang. “For example, when designing a molecule for a well-known cancer-related protein called CDK2, we found that only YuelDesign could capture the important structural changes that occur when the drug binds.”
Speed up the drug development process with AI
The research team hopes that AI tools will “democratize” drug discovery by making the drug discovery process more cost-effective and reaching patients faster.
“Our ultimate goal is to make drug discovery faster, cheaper, and more likely to succeed, bringing promising treatments to patients faster,” Dohoryan said, adding, “We have made all of our tools freely available to the scientific community. We want researchers around the world to be able to use them to tackle the diseases that matter most to patients.”
Source link
