Scientists at the University of Virginia School of Medicine announced on Apr. 9 the development of a set of artificial intelligence-powered tools designed to accelerate drug development and discovery.
The new suite, called YuelDesign, YuelPocket, and YuelBond, aims to transform how new medicines are created by improving both the design of drugs and the evaluation of existing ones for new uses. The centerpiece tool, YuelDesign, employs diffusion models—an advanced form of AI—to generate drug molecules that fit precisely into their protein targets. This method accounts for proteins' natural flexibility during binding.
YuelPocket helps pinpoint where a drug can attach to a protein using graph neural networks, while YuelBond ensures that chemical bonds in designed molecules are accurate. Together, these tools seek to address major challenges in drug development: high costs and low success rates. According to the researchers, nearly 90% of new drugs fail during human testing due largely to difficulties in predicting how well a molecule will bind with its target.
Researcher Dr. Jian Wang said: "Most existing AI tools treat the protein as a frozen statue, but that's not how biology works. Our approach lets the protein and the drug candidate evolve together during the design process, just as they would in the body." Wang added that when designing molecules for cancer-related proteins like CDK2, only YuelDesign could capture critical structural changes occurring upon binding.
Nikolay V. Dokholyan said: "Our ultimate goal is to make drug discovery faster, cheaper and more likely to succeed, so that promising treatments can reach patients sooner." He also stated his intention is "to democratize" drug discovery by making these tools freely available worldwide: "We want researchers anywhere in the world to be able to use them to tackle the diseases that matter most to their patients."
The research team includes Wang, Dong Yan Zhang, Shreshty Budakoti and Dokholyan himself. Their findings have been published in scientific journals such as PNAS, JCIM and Science Advances.
