A broadly neutralizing antibody 1E5 was optimized into a ScFv using RFDiffusion before ESMfold-suggested mutations were incorporated into designs. Candidates were selected through AlphaFold 3 and ESM predictions where binding was analyzed. Favorable candidates were further modified through rational design where the binding region with the glycoprotein was optimized for selected candidates. Rationally designed candidates were validated using ESMfold, AlphaFold 2, and AlphaFold 3 before submission, ensuring that all candidates had average pTM, pLDDT, and ipTM >80 while possessing favorable PAE. To ensure consistent binding, AlphaFold 3 was used repeatedly (50+ total recycles for each submission) to ensure that predictions possessed low variation across predictions. ESMFold and Boltz2 were also used for assessment.

As the designed ScFvs do not target the (generally) poorly predicted N-terminal hairpin helix region of the nipah virus glycoprotein, they possess low risk for off-target effects. They are likely to bind to the desired target because of the aromatic interactions between the VH of the ScFv and the nipah virus glycoprotein. It would be greatly appreciated if this set of proteins can be evaluated for effectiveness. Thank you for your support!