PDAgent: An LLM-Driven Autonomous Agent Framework Towards *In Silico* Protein Design via Directed Mutation

Song Ouyang, Zhijie Dong, Yong Luo, Kehua Su, Huangxuan Zhao, Miaojing Shi, Bo Du
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:95170-95196, 2026.

Abstract

Computational protein design holds immense promise across diverse domains, but existing approaches face significant challenges: traditional physics-based methods require substantial domain expertise, while emerging deep learning methods often rely on restricted functional ontologies, struggle to bridge the semantic gap between text and protein sequences, or lack closed-loop optimization mechanisms. In this paper, we present PDAgent, an LLM-driven autonomous agent framework that enables in silico protein design through template-based directed mutation. Our framework accepts natural language specifications of desired protein properties and employs a ReAct-style reasoning loop comprising five phases: THINK, PLAN, ACT, OBSERVE, and REFLECT. PDAgent integrates template retrieval, conservation-aware mutation strategies, and domain-specific computational tools for property optimization across seven biophysical dimensions. Experiments on 100 diverse protein design tasks demonstrate that PDAgent achieves a 91.86% average constraint satisfaction rate with high structural quality (mean pLDDT 87.69), substantially outperforming both direct LLM generation and specialized deep learning methods. We provide the source code at https://github.com/Gift-OYS/PDAgent.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-ouyang26a, title = {{PDA}gent: An {LLM}-Driven Autonomous Agent Framework Towards *{I}n Silico* Protein Design via Directed Mutation}, author = {Ouyang, Song and Dong, Zhijie and Luo, Yong and Su, Kehua and Zhao, Huangxuan and Shi, Miaojing and Du, Bo}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {95170--95196}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/ouyang26a/ouyang26a.pdf}, url = {https://proceedings.mlr.press/v306/ouyang26a.html}, abstract = {Computational protein design holds immense promise across diverse domains, but existing approaches face significant challenges: traditional physics-based methods require substantial domain expertise, while emerging deep learning methods often rely on restricted functional ontologies, struggle to bridge the semantic gap between text and protein sequences, or lack closed-loop optimization mechanisms. In this paper, we present PDAgent, an LLM-driven autonomous agent framework that enables in silico protein design through template-based directed mutation. Our framework accepts natural language specifications of desired protein properties and employs a ReAct-style reasoning loop comprising five phases: THINK, PLAN, ACT, OBSERVE, and REFLECT. PDAgent integrates template retrieval, conservation-aware mutation strategies, and domain-specific computational tools for property optimization across seven biophysical dimensions. Experiments on 100 diverse protein design tasks demonstrate that PDAgent achieves a 91.86% average constraint satisfaction rate with high structural quality (mean pLDDT 87.69), substantially outperforming both direct LLM generation and specialized deep learning methods. We provide the source code at https://github.com/Gift-OYS/PDAgent.} }
Endnote
%0 Conference Paper %T PDAgent: An LLM-Driven Autonomous Agent Framework Towards *In Silico* Protein Design via Directed Mutation %A Song Ouyang %A Zhijie Dong %A Yong Luo %A Kehua Su %A Huangxuan Zhao %A Miaojing Shi %A Bo Du %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-ouyang26a %I PMLR %P 95170--95196 %U https://proceedings.mlr.press/v306/ouyang26a.html %V 306 %X Computational protein design holds immense promise across diverse domains, but existing approaches face significant challenges: traditional physics-based methods require substantial domain expertise, while emerging deep learning methods often rely on restricted functional ontologies, struggle to bridge the semantic gap between text and protein sequences, or lack closed-loop optimization mechanisms. In this paper, we present PDAgent, an LLM-driven autonomous agent framework that enables in silico protein design through template-based directed mutation. Our framework accepts natural language specifications of desired protein properties and employs a ReAct-style reasoning loop comprising five phases: THINK, PLAN, ACT, OBSERVE, and REFLECT. PDAgent integrates template retrieval, conservation-aware mutation strategies, and domain-specific computational tools for property optimization across seven biophysical dimensions. Experiments on 100 diverse protein design tasks demonstrate that PDAgent achieves a 91.86% average constraint satisfaction rate with high structural quality (mean pLDDT 87.69), substantially outperforming both direct LLM generation and specialized deep learning methods. We provide the source code at https://github.com/Gift-OYS/PDAgent.
APA
Ouyang, S., Dong, Z., Luo, Y., Su, K., Zhao, H., Shi, M. & Du, B.. (2026). PDAgent: An LLM-Driven Autonomous Agent Framework Towards *In Silico* Protein Design via Directed Mutation. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:95170-95196 Available from https://proceedings.mlr.press/v306/ouyang26a.html.

Related Material