ACCELERATING ANTIBODY DESIGN WITH ACTIVE LEARNING
S.-w. Seo, M. W. Kwak, E. Kang, C. Kim, E. Park, T. H. Kang, J. Kim
The paper develops a sequence generation model using transfer and active learning to efficiently discover binding antibodies with novel HCDR3 regions, requiring only six thousand data points per antigen.
bioRxiv · Sep 12, 2022 · PreprintBioengineering