Accuracy of lowest-energy docking decoy for different docking protocols.

All simulations were carried out using the lowest-energy RosettaAntibody homology model, except (1) the Crystal column where the crystal structure of the antibody was used, and (2) the last two columns where the ensemble of the ten lowest-energy RosettaAntibody homology models were used as EnsembleDock starting structures. Type indicates bound (B) or unbound (U) starting structures for docking for the antibody (Ab) or antigen (Ag). The corresponding PDB IDs of the unbound structures are noted in parentheses. The number of residues in the CDR H3 loop is provided since it is at the center of the paratope and is a key contributor to accurate antibody-antigen interactions. Docking performance is indicated using CAPRI ratings [34]: Three stars (***), two stars (**), one star (*) and “0” indicate that lowest-energy docking decoy is of high, medium, acceptable and incorrect quality, respectively. Predictions marked “f” additionally have a docking energy funnel defined where five of the ten lowest-energy structures are of medium or high quality. CAPRI summary lines show the total number of targets for which the lowest-energy docking decoy and the most accurate model in the ten lowest-energy docking decoys is of high/medium/acceptable quality. Full quantitative measurements underlying the CAPRI ratings, of the predicted models are available in Supporting Information Table S1.