AI finds a weak switch in herpes viruses-and shuts entry down cold
Date: December 15, 2025
Source: Washington State University
Washington State University researchers used AI, simulations, and lab experiments to expose a single amino acid interaction that herpes viruses need to enter cells. Changing that one residue stopped viral fusion with cells altogether. The study appears in the journal Nanoscale and outlines a fast, focused path for antiviral discovery.
Why this matters for antiviral R&D
Herpes viruses depend on a large "fusion" protein to merge with cell membranes. The protein's shape-shifting is hard to map and even harder to disrupt, which is one reason vaccine progress has lagged. Targeting a specific interaction inside this protein gives researchers a precise lever to test-rather than sifting through thousands of guesses.
How the team found the weak spot
Using an algorithm to map amino-acid interactions, the team applied machine learning to rank which contacts mattered most for entry. From thousands of possibilities, one interaction stood out as indispensable. Professors Prashanta Dutta and Jin Liu led the computational work across the School of Mechanical and Materials Engineering.
In follow-up experiments led by Anthony Nicola in Veterinary Microbiology and Pathology, the group introduced a targeted mutation at that key site. Result: the virus failed to fuse with cells and could not enter. As Liu put it, "Most interactions are background noise, but a few are critical-and finding them with computation saves years of trial and error."
Key findings
- A single amino acid interaction in the herpes fusion protein is essential for entry.
- Mutating that residue blocked fusion and stopped infection in cell-based experiments.
- AI-guided filtering compressed what could take years into a tractable, testable shortlist.
What this enables next
- Rational antiviral design: Small molecules, peptides, or antibodies can be aimed at the specific interaction that gates fusion.
- Generalizable workflow: The same AI-simulation-experiment loop can probe other viral fusion systems.
- Faster iteration: Computational triage narrows hypotheses before committing months of bench time.
Caveats and open questions
The study pinpoints an essential interaction, but the full structural ripple effect across the fusion protein still needs to be mapped. Bridging what simulations "see" with what experiments measure at larger scales remains a challenge. Further work will test how this mechanism behaves across strains and in more complex models.
Project details
- Institutions: School of Mechanical and Materials Engineering; Department of Veterinary Microbiology and Pathology, Washington State University.
- Authors include Jin Liu, Prashanta Dutta, Anthony Nicola, and PhD students Ryan Odstrcil, Albina Makio, and McKenna Hull.
- Funding: National Institutes of Health (NIH).
- Publication: Nanoscale (Royal Society of Chemistry). See journal info: Nanoscale.
For teams building antiviral pipelines
- Integrate MD simulations early to score residue-residue interactions in fusion proteins.
- Use ML ranking to prioritize a short list of contacts for mutational testing.
- Design assays that read out fusion directly (e.g., cell-cell fusion, pseudovirus entry).
- Iterate: fold new structural data back into the model to refine targets.
For broader context on antiviral strategies, see NIAID: Antiviral Drugs.
Upskilling your lab's AI capability
If you're formalizing AI skills for your research group-from model selection to experiment triage-browse these curated options by role: AI courses by job.
Your membership also unlocks:
