AbAgKer: a unified semi-supervised framework for antigen-antibody binding affinity and kinetics prediction
Gang Luo, Junkai Wang, Sizhe Zhang, Zhilin Zhu, Zhangli Lu, Min Li, Pier Luigi Martelli · 2026
Read AbAgKer: a unified semi-supervised framework for antigen-antibody binding affinity and kinetics prediction by Gang Luo, Junkai Wang, Sizhe Zhang, Zhilin Zhu, Zhangli Lu, Min Li, Pier Luigi Martelli (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13516917.txt.
Description
Luo and co-authors present AbAgKer, a computer model for judging antibody candidates, in this 2026 paper. New design tools can propose large numbers of therapeutic antibodies, but checking how tightly each one grips its target antigen still depends on scarce lab data. The model predicts both binding strength and binding kinetics, meaning how fast an antibody attaches and lets go.
AbAgKer blends learned sequence features with biological knowledge. That knowledge covers the antigen's binding region and the antibody's variable loops. A mixture-of-experts design routes data to specialized sub-models. To cope with limited labeled examples, it learns partly from its own predictions on unlabeled data, a strategy called self-distillation. In tests it beat comparison models, and its learned representations transferred to predicting dissociation rates, which govern how long a drug stays attached.
Catalog record
| CREATOR | Gang Luo, Junkai Wang, Sizhe Zhang, Zhilin Zhu, Zhangli Lu, Min Li, Pier Luigi Martelli |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |