Cambridge AI designs first fully machine-generated vaccine antigen

University of Cambridge researchers used a protein language model trained on 2.3 million sequences to generate a stabilized prefusion SARS-CoV-2 spike fragment. The candidate passed a 42-person Phase I safety trial with zero grade-3 adverse events and induced neutralizing titers above 1:640 in 38 participants after two doses. The entire design-to-clinic timeline was 18 months instead of the typical 4 years.

You treat antigen selection as an optimization loop rather than a manual literature search. The technique shows that sequence-to-structure models can compress early discovery from years to months, so you allocate budget to rapid iteration instead of exhaustive wet-lab screening. This changes prioritization from "which candidate looks safest" to "which candidate scores highest on the model and clears the first animal screen."

The Cambridge Centre for AI in Medicine released the model weights and trial protocol under an open license; their generated antigen is now advancing to a 500-person Phase II study funded by the UK Vaccines Taskforce. The group reports a 60 percent drop in candidate rejection rate compared with their prior manual pipeline.