Multiproperty optimization of therapeutic antibodies through sequence editing via large language model and reinforcement learning
Abstract
Discovering antibodies with multiple beneficial properties is essential for therapeutic drug development, but experimentally identified antibodies require further optimization. Joint optimization remains challenging because sequence space is vast, annotations are sparse and property tradeoffs are common. Herein, we propose AbOptimizer, a reinforcement learning framework for multiproperty antibody optimization through parallel multisite sequence editing. AbOptimizer identifies promising editing sites and edit actions, including ‘reservation’, ‘substitution’ and optional ‘insertion’ or ‘deletion’, before recommending mutations from protein language models, while surrogate evaluation and Pareto rewards shape policy training. We evaluate AbOptimizer on three therapeutic antibodies across multiproperty, data-limited and indel-inclusive optimization. In silico, AbOptimizer improves optimized properties and outperforms baseline methods. In vitro, selected variants show affinity-specificity co-optimization for emibetuzumab, improved affinity and specificity with largely maintained thermal stability for siltuximab, and indel-inclusive affinity optimization for trastuzumab. These results validate AbOptimizer as a practical framework for therapeutic antibody optimization.
Related
- Targeting Integrins in Triple Negative Breast Cancer: Therapeutic Potential of Monoclonal Antibodies, PARP Inhibitors, Synthetic Peptides, and Aptamers
- A Multiplex Protein Microarray Platform for Preclinical Tissue Binding and Cross-Reactivity of Therapeutic Antibodies
- Retail sales optimization using a hybrid K-Means, particle swarm optimization and reinforcement learning framework
- Adaptive intelligent resource optimization in wireless sensor networks using reinforcement learning and compressive sensing
- Continuous control deep reinforcement learning for joint resource optimization in STAR-RIS assisted mmWave-NOMA systems with hardware impairments
BibTeX
@article{qiu2026multiproperty,
title = {{Multiproperty optimization of therapeutic antibodies through sequence editing via large language model and reinforcement learning}},
author = {Minyao Qiu and Zhiwei Wang and Xuan Liu and Hongzhun Wang and Quangao Hua and Qin Shu and Shaoxiang Luo and Mengxi Niu and Feng Ma and Feng Huang and Yongkang Wang and Zhankun Xiong and Wen Zhang},
journal = {Nature Communications},
year = {2026},
month = oct,
doi = {10.1038/s41467-026-78533-6},
url = {https://www.nature.com/articles/s41467-026-78533-6},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Multiproperty optimization of therapeutic antibodies through sequence editing via large language model and reinforcement learning
AU - Minyao Qiu
AU - Zhiwei Wang
AU - Xuan Liu
AU - Hongzhun Wang
AU - Quangao Hua
AU - Qin Shu
AU - Shaoxiang Luo
AU - Mengxi Niu
AU - Feng Ma
AU - Feng Huang
AU - Yongkang Wang
AU - Zhankun Xiong
AU - Wen Zhang
PY - 2026
DA - 2026/10/26
JO - Nature Communications
DO - 10.1038/s41467-026-78533-6
UR - https://www.nature.com/articles/s41467-026-78533-6
AB - Discovering antibodies with multiple beneficial properties is essential for therapeutic drug development, but experimentally identified antibodies require further optimization. Joint optimization remains challenging because sequence space is vast, annotations are sparse and property tradeoffs are common. Herein, we propose AbOptimizer, a reinforcement learning framework for multiproperty antibody optimization through parallel multisite sequence editing. AbOptimizer identifies promising editing sites and edit actions, including ‘reservation’, ‘substitution’ and optional ‘insertion’ or ‘deletion’, before recommending mutations from protein language models, while surrogate evaluation and Pareto rewards shape policy training. We evaluate AbOptimizer on three therapeutic antibodies across multiproperty, data-limited and indel-inclusive optimization. In silico, AbOptimizer improves optimized properties and outperforms baseline methods. In vitro, selected variants show affinity-specificity co-optimization for emibetuzumab, improved affinity and specificity with largely maintained thermal stability for siltuximab, and indel-inclusive affinity optimization for trastuzumab. These results validate AbOptimizer as a practical framework for therapeutic antibody optimization.
SN - 2041-1723
ER -