← Back to Home

Protein Design Moves Beyond Model Scores: London Meeting Focuses on Experiments and Methodological Limitations

From antiviral antibodies to plastic-degrading enzymes, protein design has an increasingly diverse range of applications. The inaugural London Protein Design Day brought researchers together to discuss how to determine whether computer-generated candidate molecules can actually perform their intended tasks.

By SURL BioNews

Computers can propose new protein sequences, but whether they can block viruses, recognize specific molecules or break down plastic ultimately requires experimental answers. The inaugural London Protein Design Day placed this question at the center of its discussions: as design tools rapidly proliferate, researchers need to share more than model scores, including the limitations methods encounter in practice.

The event took place on June 23, 2026, at Imperial College London. In a recap published in BioTechniques on November 6, organizer Aubin Ramon summarized the research and discussions presented that day. The official event page stated that the symposium was organized by the Sormanni lab, with venue support from Imperial College London’s Department of Chemical Engineering and sponsorship from Boltz Bio. Topics covered AI design, molecular dynamics and bioinformatics, as well as applications involving enzymes, antibodies and materials.

In biomedical research, speakers presented approaches that combine modeling of antibody–antigen interfaces with antibody redesign to address SARS-CoV-2 variants. Other research used protein engineering to explore inhibitors of Zika and dengue viruses, diagnostic tools and self-assembling vaccine platforms. Each of these topics has different validation requirements: binding to a target is only one step. Whether a design can block infection, distinguish closely related viruses or elicit an appropriate immune response requires corresponding functional experiments.

Enzyme and materials research offered another measure of what it means to be effective. Ramon’s recap described a team combining searches of natural enzyme diversity with evolutionary information to identify heat-resistant enzymes suitable for PET plastic recycling. Other work involved designing proteins around cofactors to explore components for electron and energy transfer. These applications remind researchers that design quality must be assessed in the intended operating environment and cannot be captured by a single set of computational metrics.

The discussions also addressed attempts that had not yet succeeded. According to Ramon, some speakers shared preliminary results openly to seek feedback from peers and discuss which evaluation metrics are meaningful, as well as how the same tool performs when the target or operator changes. However, the event recap did not provide complete datasets, control designs or clinical results that could be checked item by item. These presentations are therefore best viewed as research directions and practical experience; they do not establish that the therapies or vaccines concerned are effective.

The gathering also reflected demand within the UK research community for spaces to exchange ideas. In a public post after the event, Ramon said that nearly 300 people from academia, industry, venture capital and government had registered, but the venue could not accommodate everyone. The organizing team consequently planned a larger space for the next edition. For protein design, the value of such gatherings lies in enabling researchers across different fields to bring their experimental experience back into tool development, gradually clarifying which designs are truly useful.

References

  1. BioTechniques
  2. Luma — event organizers
  3. Aubin Ramon / LinkedIn