Protein as a Second Language for LLMs

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chen, Xinhui, Li, Zuchao, Gao, Mengqi, Zhang, Yufeng, Leong, Chak Tou, Li, Haoyang, Chen, Jiaqi
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909841358848000
author Chen, Xinhui
Li, Zuchao
Gao, Mengqi
Zhang, Yufeng
Leong, Chak Tou
Li, Haoyang
Chen, Jiaqi
author_facet Chen, Xinhui
Li, Zuchao
Gao, Mengqi
Zhang, Yufeng
Leong, Chak Tou
Li, Haoyang
Chen, Jiaqi
contents Deciphering the function of unseen protein sequences is a fundamental challenge with broad scientific impact, yet most existing methods depend on task-specific adapters or large-scale supervised fine-tuning. We introduce the "Protein-as-Second-Language" framework, which reformulates amino-acid sequences as sentences in a novel symbolic language that large language models can interpret through contextual exemplars. Our approach adaptively constructs sequence-question-answer triples that reveal functional cues in a zero-shot setting, without any further training. To support this process, we curate a bilingual corpus of 79,926 protein-QA instances spanning attribute prediction, descriptive understanding, and extended reasoning. Empirically, our method delivers consistent gains across diverse open-source LLMs and GPT-4, achieving up to 17.2% ROUGE-L improvement (average +7%) and even surpassing fine-tuned protein-specific language models. These results highlight that generic LLMs, when guided with protein-as-language cues, can outperform domain-specialized models, offering a scalable pathway for protein understanding in foundation models.
format Preprint
id arxiv_https___arxiv_org_abs_2510_11188
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Protein as a Second Language for LLMs
Chen, Xinhui
Li, Zuchao
Gao, Mengqi
Zhang, Yufeng
Leong, Chak Tou
Li, Haoyang
Chen, Jiaqi
Machine Learning
Artificial Intelligence
Biomolecules
Deciphering the function of unseen protein sequences is a fundamental challenge with broad scientific impact, yet most existing methods depend on task-specific adapters or large-scale supervised fine-tuning. We introduce the "Protein-as-Second-Language" framework, which reformulates amino-acid sequences as sentences in a novel symbolic language that large language models can interpret through contextual exemplars. Our approach adaptively constructs sequence-question-answer triples that reveal functional cues in a zero-shot setting, without any further training. To support this process, we curate a bilingual corpus of 79,926 protein-QA instances spanning attribute prediction, descriptive understanding, and extended reasoning. Empirically, our method delivers consistent gains across diverse open-source LLMs and GPT-4, achieving up to 17.2% ROUGE-L improvement (average +7%) and even surpassing fine-tuned protein-specific language models. These results highlight that generic LLMs, when guided with protein-as-language cues, can outperform domain-specialized models, offering a scalable pathway for protein understanding in foundation models.
title Protein as a Second Language for LLMs
topic Machine Learning
Artificial Intelligence
Biomolecules
url https://arxiv.org/abs/2510.11188