Researchers at Texas A&M announced on July 30 the development of artificial intelligence tools aimed at improving the drug discovery process for tuberculosis. The team, led by James Sacchettini, Ph.D., has created systems to help scientists focus their efforts after initial screening and organize years of collaborative data into a searchable format.
Sacchettini, who holds the Rodger J. Wolfe-Welch Foundation Chair in Science and is a professor in multiple departments at Texas A&M, said his lab's work with AI addresses a significant challenge in tuberculosis research: narrowing down thousands of potential compounds identified during early drug screens. "We might get thousands of compounds from a screen and then have to decide, which one are we going to work on?" Sacchettini said.
The lab developed DAIKON, an open-source platform published in 2023 that tracks drug targets from gene identification through years of chemistry work. The Tuberculosis Drug Accelerator (TBDA), supported by the Gates Foundation, uses DAIKON across its partnership network. New AI tools developed by Sacchettini's team integrate directly with this platform.
One key tool is CAGE-Fusion, an AI model trained on published screening data to identify problematic compounds known as 'nuisance molecules.' These can give false signals during testing and lead researchers down costly paths. According to Siddhant Rath, an AgriLife Research scientist who led the effort, "The model can walk you through the process and show you which regions of the molecules it thought were problematic." The model successfully flags suspicious compounds about 94% of the time when comparing known nuisance molecules against clean ones.
Another system developed by Rath and Saswati Panda helps TBDA partners navigate large datasets more efficiently. This tool enables users to visually trace molecular structures across projects and quickly access related presentations or documentation through a chat interface.
Sacchettini said these advances are not expected to provide definitive answers but serve as valuable guides for researchers: "We're not hoping for AI to give us the exact right answer," he said. "But it can tell us what not to work on, which then informs us what we should be working on. And it really is a big time saver."
