UCF scientist using AI to speed up TB drug discovery
Using Kyle Rohde’s data and artificial intelligence, biotech company Atomwise narrowed its focus from millions of drugs to just 100, accelerating the search for new tuberculosis antibiotics.
As scientists harness artificial intelligence (AI) to accelerate drug discovery, UCF College of Medicine tuberculosis researcher Kyle Rohde has partnered with biotech company Atomwise to see how AI can speed up testing of potential drugs to treat the disease.
AI can hasten scientific research by quickly testing potentially millions of drugs against a disease model.
Tuberculosis, or TB, cases rose 16% from 2022 to 2023, with more than 9,600 cases reported in the U.S., according to a study by the U.S. Centers for Disease Control and Prevention. The rise of multi-drug resistant strains of Mycobacterium tuberculosis is also a public health concern.
For those reasons, scientists like Rohde are looking for ways to speed up discovery of drugs to better tuberculosis.
Atomwise contacted Rohde and researchers across 30 countries worldwide to participate in using an artificial intelligence molecular screen (AIMS) to find new effective TB drugs. Atomwise is a San Francisco-based technology-enabled pharmaceutical company. Its goal is “to invent a better way to discover drugs that help patients,” according to its website.
Rohde agrees that technology will play a key role in speeding up drug discovery.
“Traditionally, drug discovery has involved high-throughput screening (HTS) of hundreds of thousands to millions of compounds to find hits that inhibit a target protein or kill the bacteria,” he says. “This can entail millions of dollars and years of testing just to find the starting point for a few antibiotic candidates.”
Instead, using data supplied by Rohde and the other worldwide scientists, Atomwise developed a proprietary AI model that can predict inhibitors of selected targets, in this case two proteins critical for TB survival. Using this tool, Atomwise screened 2.5 million commercially available compounds to identify potential new antibiotics for TB.
“With the help of AI, Atomwise was able to eliminate more than two million compounds unlikely to inhibit the TB bacteria,” Rohde says. “That narrows our focus to less than 100 top candidates. Testing 100 drugs versus millions will undoubtably accelerate our research, saving a lot of time and money.”
Tuberculosis, which attacks the lungs and is easily spread through the air, affects millions of people each year. If not treated properly, it can be fatal. Rohde sees AI as a valuable tool in the evolution of scientific research to keep up with drug-resistant infections.
“If successful, the use of AI-based virtual drug screening could revolutionize the challenging task of discovering new antibiotics,” he says. “It has the potential to really accelerate the development of affordable new treatment options for global health problems like TB.”
Source: University of Central Florida
