Interleaved Tool-Call Reasoning for Protein Function Understanding
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914369252622336 |
|---|---|
| author | Fan, Chuanliu Ma, Zicheng Meng, Huanran Zhang, Aijia Du, Wenjie Zhang, Jun Gao, Yi Qin Cao, Ziqiang Fu, Guohong |
| author_facet | Fan, Chuanliu Ma, Zicheng Meng, Huanran Zhang, Aijia Du, Wenjie Zhang, Jun Gao, Yi Qin Cao, Ziqiang Fu, Guohong |
| contents | Recent advances in large language models (LLMs) have highlighted the effectiveness of chain-of-thought reasoning in symbolic domains such as mathematics and programming. However, our study shows that directly transferring such text-based reasoning paradigms to protein function understanding is ineffective: reinforcement learning mainly amplifies superficial keyword patterns while failing to introduce new biological knowledge, resulting in limited generalization. We argue that protein function prediction is a knowledge-intensive scientific task that fundamentally relies on external biological priors and computational tools rather than purely internal reasoning. To address this gap, we propose PFUA, a tool-augmented protein reasoning agent that unifies problem decomposition, tool invocation, and grounded answer generation. Instead of relying on long unconstrained reasoning traces, PFUA integrates domain-specific tools to produce verifiable intermediate evidence. Experiments on four benchmarks demonstrate that PFUA consistently outperforms text-only reasoning models with an average performance improvement of 103%. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_03604 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Interleaved Tool-Call Reasoning for Protein Function Understanding Fan, Chuanliu Ma, Zicheng Meng, Huanran Zhang, Aijia Du, Wenjie Zhang, Jun Gao, Yi Qin Cao, Ziqiang Fu, Guohong Artificial Intelligence Recent advances in large language models (LLMs) have highlighted the effectiveness of chain-of-thought reasoning in symbolic domains such as mathematics and programming. However, our study shows that directly transferring such text-based reasoning paradigms to protein function understanding is ineffective: reinforcement learning mainly amplifies superficial keyword patterns while failing to introduce new biological knowledge, resulting in limited generalization. We argue that protein function prediction is a knowledge-intensive scientific task that fundamentally relies on external biological priors and computational tools rather than purely internal reasoning. To address this gap, we propose PFUA, a tool-augmented protein reasoning agent that unifies problem decomposition, tool invocation, and grounded answer generation. Instead of relying on long unconstrained reasoning traces, PFUA integrates domain-specific tools to produce verifiable intermediate evidence. Experiments on four benchmarks demonstrate that PFUA consistently outperforms text-only reasoning models with an average performance improvement of 103%. |
| title | Interleaved Tool-Call Reasoning for Protein Function Understanding |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2601.03604 |