Researchers at the University of Texas Medical Branch in Galveston have built an artificial intelligence pipeline that sifts through alphavirus genomes to identify the protein fragments most likely to trigger a strong immune response, a step toward vaccines that could protect against a family of mosquito-borne viruses for which almost no licensed shots exist.
The effort, led by virologist Nikos Vasilakis and UTMB Chief AI Officer Peter McCaffrey, comes at a precarious moment for alphavirus vaccination. Chikungunya remains the only alphavirus for which the U.S. Food and Drug Administration has ever approved a vaccine, and that product, Ixchiq, is now sidelined. The FDA suspended Ixchiq’s biologics license after a review of safety data, leaving the U.S. without a commercially available alphavirus vaccine and underscoring the need for new approaches.
Alphaviruses circulate on every inhabited continent. Chikungunya alone has caused millions of infections across the Americas, Africa and Asia, leaving many patients with debilitating joint pain that can persist for months or years. Other members of the family, including eastern equine encephalitis virus and Venezuelan equine encephalitis virus, can invade the brain and carry case-fatality rates above 30 percent. No broadly protective vaccine covers more than one of these threats.
How the AI pipeline works
The UTMB system uses machine learning to predict epitopes, the short stretches of viral protein that immune cells latch onto. A peer-reviewed study published in PMC describes the full workflow: algorithms scan alphavirus protein sequences, score candidate epitopes for predicted immunogenicity and coverage across diverse human leukocyte antigen (HLA) types, then feed the top hits into protein modeling and molecular docking software. Candidates are ranked by a composite of immunogenicity, HLA allele coverage, solubility and structural stability.
The pipeline is designed to be repeatable. When a new alphavirus genome is sequenced, the same process can generate a fresh set of ranked targets without starting from scratch, an advantage over traditional trial-and-error epitope selection.
Lab validation backs up the predictions
Crucially, the AI-selected epitopes have not stayed on a hard drive. The same PMC study reports that the team tested predicted peptides on microarrays and confirmed receptor binding in follow-up experiments. A UTMB news release from April 2026 adds that the group has measured T-cell activation and the release of key cytokines, interferon-gamma, TNF-alpha and interleukin-2, in response to selected epitopes. Those markers are standard benchmarks vaccine developers use to gauge whether a candidate can stimulate cellular immunity.
Independent research supports the biological logic behind the approach. A 2021 study in Cell showed that human monoclonal antibodies can recognize a single conserved site on the alphavirus E1 protein and protect animals against both arthritis-causing and brain-invading alphaviruses. That finding suggests a shared vulnerability across the family, exactly the kind of target the UTMB pipeline is designed to find.
Infrastructure for dangerous work
Testing vaccine concepts against viruses like eastern equine encephalitis requires high-containment laboratories. UTMB houses the Galveston National Laboratory, which the National Institute of Allergy and Infectious Diseases describes as part of the NIH biocontainment network, with capacity for work on some of the world’s most dangerous pathogens. That infrastructure gives the team a path from computer-predicted epitopes to live-virus challenge studies without shipping materials to an outside facility.
McCaffrey’s dual role reflects UTMB’s broader institutional bet on AI. The university formally established an Artificial Intelligence Center in May 2025 and appointed McCaffrey as its director and the institution’s first Chief AI Officer, positioning vaccine design as one application within a campus-wide AI strategy.
What remains uncertain
For all the progress in silico and in vitro, several large questions are unanswered.
No animal protection data yet. The published pipeline paper documents peptide binding and receptor interaction, not whether vaccinated animals resist infection. The UTMB news release describes T-cell activation but stops short of reporting challenge studies. By contrast, the 2021 Cell study on naturally derived antibodies did demonstrate in vivo protection, a bar the AI-designed candidates have not yet publicly cleared.
Funding is undisclosed. None of the available sources identify who is paying for the pipeline work, whether it is NIH grants, Department of Defense contracts, internal UTMB funds or industry partnerships. That gap makes it difficult to assess the project’s scale or timeline.
Regulatory path is undefined. The FDA’s original approval of Ixchiq relied on immune-response endpoints rather than direct efficacy data, showing the agency is open to surrogate measures for alphavirus vaccines. But no public guidance addresses how regulators would evaluate epitopes chosen by machine learning versus traditional methods. The CDC’s current chikungunya vaccine guidance is built around licensed products and specific risk groups; it does not contemplate AI-designed candidates.
External partnerships are unclear. NIAID recognizes the Galveston National Laboratory as part of its network, and UTMB names Vasilakis and McCaffrey as leads, but no source specifies whether outside agencies, pharmaceutical companies or international collaborators are formally involved in the epitope pipeline.
Why it matters now
The suspension of Ixchiq’s license has left a practical gap. Chikungunya continues to circulate in tropical and subtropical regions, and the CDC had recommended the vaccine for travelers and laboratory workers in certain risk categories. With that product unavailable, there is no licensed alphavirus vaccine on the U.S. market as of May 2026.
Meanwhile, climate change and urbanization are expanding the geographic range of Aedes mosquitoes that carry chikungunya, Mayaro and other alphaviruses. Eastern equine encephalitis, though rare, has appeared in new U.S. counties in recent years, and its high fatality rate makes even small outbreaks alarming.
The UTMB pipeline does not solve these problems today. What it offers is a systematic, reproducible method for generating vaccine targets across an entire virus family, one that has already produced epitopes capable of activating human immune cells in the lab. Whether those targets can protect people will depend on animal studies, manufacturing decisions and regulatory reviews that have not yet begun in public view. But with the only licensed alphavirus vaccine now off the market, the pressure to find alternatives is real and growing.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
