UTMB Researchers Develop AI Pipeline to Accelerate Pan-Alphavirus Vaccine Development

A team of scientists at The University of Texas Medical Branch (UTMB), led by Nikos Vasilakis, PhD, and Peter McCaffrey, MD, has developed a new computational pipeline that could dramatically accelerate the development of vaccines against a group of mosquito-borne viruses known as alphavirus.

Vasilakis is a professor and the vice chair for research, and McCaffrey is an assistant professor of clinical practice and director of the UTMB AI center, both in the Department of Pathology. The work was conducted in collaboration with the researchers' colleagues in Brazil and Panama.

The research, published in the journal Science Advances, describes an integrated reiterative vaccine design system that uses machine learning, structural biology, and laboratory validation to identify promising vaccine targets across multiple related viruses simultaneously.

Alphaviruses, which cause outbreaks of diseases such as chikungunya and equine encephalitis, are transmitted by mosquitoes and can cause severe fever, arthritis, and neurological disease in humans and animals.

Alphaviruses continue to emerge and reemerge worldwide, and traditional vaccine development simply cannot keep pace."

Nikos Vasilakis, PhD, UTMB

To address this challenge, researchers designed a rapid "pipeline" that analyzes viral proteins to identify short fragments, called epitopes, that can trigger a strong immune response. The system evaluates these fragments based on factors such as immunogenicity, genetic coverage across populations, stability, and solubility.

"Our pipeline offers a new way forward by integrating computational prediction with experimental validation to identify vaccine targets that could provide broad protection across multiple viruses at once," Vasilakis said.

The researchers also validated their predictions using peptide microarrays and molecular modeling to confirm that the selected epitopes bind appropriately to immune receptors. These analyses helped identify dozens of epitopes that appear reactive across multiple alphavirus species - an important step toward creating a pan-alphavirus vaccine that could protect against several viruses at once. 

Using the pipeline, scientists screened hundreds of viral peptides and identified a set of candidate epitopes capable of stimulating immune responses across multiple alphaviruses. Laboratory tests using immune cells from both mice and humans showed that several of these peptides activated T cells and triggered the release of key immune signaling molecules such as interferon-gamma, tumor necrosis factor-alpha, and interleukin-2. 

"Instead of approaching vaccine development one virus at a time, this platform allows us to think more strategically and proactively and to perform analytical work at a scale that previously was not achievable," McCaffrey said. "By combining machine learning, structural biology, and laboratory testing, we can rapidly narrow down the most promising targets and accelerate the path toward broadly protective vaccines."

In addition to identifying potential vaccine targets, the pipeline establishes a repeatable workflow that integrates computational prediction with experimental validation - providing a template for future vaccine development efforts.

Researchers are now continuing to evaluate the most promising vaccine candidates in animal models, with the goal of advancing broadly protective vaccines against alphaviruses and other emerging pathogens.

"This work is the first publication that utilized artificial intelligence and machine learning to develop a pan-genus vaccine candidate that has been validated experimentally," Vasilakis said.