Summary
ESM-IF outperforms ProteinMPNN on designing high-affinity structures of antibodies in both a zero-shot (1) and few-shot (2) setting. In other settings, ProteinMPNN showed poor-to-modest correlation with antigen binding(3).
Figures
Ref (2)
Ref (1)
Ref (3)
1.
Shanker VR, Bruun TUJ, Hie BL, Kim PS. Unsupervised evolution of protein and antibody complexes with a structure-informed language model. Science. 2024;385(6704):46–53. Available from: https://doi.org/10.1126/science.adk8946
2.
Hoie M, Hummer A, Olsen T, Nielsen M, Deane C. AntiFold: Improved antibody structure design using inverse folding. In: GenBio@NeurIPS2023. 2023. Available from: https://openreview.net/forum?id=bxZMKHtlL6
3.
Chun C, Sohn B-K, Ki H, Jo JH, An H-T, Park J, et al. Landscape-scale navigation unlocks antibody CDR structural logic for AI-guided rescue and therapeutic optimization. 2026 Apr; Available from: http://dx.doi.org/10.64898/2026.04.21.719857