Results of common germline gene multi-specificity design benchmark.

Three sets of antibodies encoded by a common germline gene were designed against their targets using RECON, both by fixed backbone (FBB) and backbone minimized (BBM) designs, and MPI_MSD algorithms. Designs were evaluated by recovery of the germline sequence and the fitness of designed models.
aFitness is defined as the sum of the Rosetta score of all input complexes, in Rosetta Energy Units (REU). Fitness was reported for the top ten complexes generated by each design method.
Results of common germline gene multi-specificity design benchmark.