First fully AI-designed antigen passes early human safety test
University of Cambridge researchers used a generative diffusion model to design a stabilized pre-fusion SARS-CoV-2 spike epitope, synthesized the 38-residue peptide, and formulated it with an adjuvant. In a Phase 1 trial of 32 volunteers the candidate produced neutralizing titers above 1:640 with no serious adverse events.
Biologists see that protein language models can now propose viable immunogens in days rather than months. Labs should integrate these models into their antigen discovery pipelines to shorten early candidate selection.
Cambridge's Scherman Group released the epitope sequence under open license; VaxEquity licensed it and reported the same sequence elicited cross-variant protection in ferret challenge studies with 92 percent survival.
Step 1: Go to https://github.com/cambridge-ai-bio/diffusion-antigen and install the conda environment. Step 2: Provide your target pathogen FASTA file and run `python generate.py --length 40 --temperature 0.8`. Step 3: Rank the top three sequences by the model's predicted binding energy, then order the peptides from a synthesis service for initial ELISA validation.