Figure 1.

The workflow of OptMAVEn-2.0. First, the initial epitope positioning step rotates the antigen such that its epitope points downward with epitope centroid at the origin. The grid search step generates an ensemble of antigen positions, followed by the mixed-integer linear programming (MILP) step, where the six lowest interaction energy Modular Antibody Parts (MAPs) are chosen to construct the variable antibody fragment. A Euclidean coordinate for each part in the MAPs database was generated using the embedder module. The k-means protocol uses these and the epitope centroid coordinates and rotation angle to cluster the antibodies. The antibodies with the most negative MILP energy in each cluster are then subjected to structural relaxation and a short molecular dynamics (MD) routine to verify their high affinities. Stable designs emerging from this step could be affinity matured with the dual objective of enhancing their antigen-antibody affinities and lowering their immunogenic potentials.