arXiv 2412.116182024-12-16cs.LGcs.AI

EvoLlama：通过多模态结构与序列表示增强大语言模型对蛋白质的理解

EvoLlama: Enhancing LLMs' Understanding of Proteins via Multimodal Structure and Sequence Representations

School of Computer Science and Technology, East China Normal University（华东师范大学计算机科学与技术学院）；Institute of Artificial Intelligence (TeleAI), China Telecom（中国电信人工智能研究院（TeleAI））；School of Computer Science, Fudan University（复旦大学计算机学院）；Xiaohongshu Inc（小红书科技有限公司）；Innovation Center for AI and Drug Discovery, School of Pharmacy, East China Normal University（华东师范大学药学院人工智能与药物发现创新中心）

Nuowei Liu, Changzhi Sun, Tao Ji, Junfeng Tian, Jianxin Tang, Yuanbin Wu, Man Lan

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出EvoLlama多模态框架，融合结构编码器ProteinMPNN、序列编码器ESM-2与Llama-3，通过微调显著提升LLM蛋白质理解能力，零样本性能平均提升1%-8%，超越监督微调基线6%。

摘要：

当前用于理解蛋白质的大型语言模型（LLMs）主要将氨基酸序列视为一种文本模态。与此同时，蛋白质语言模型（PLMs），如ESM-2，已经从自然蛋白质序列的宇宙中学习到了大量的序列进化知识。此外，基于结构的编码器如ProteinMPNN通过图神经网络学习蛋白质的结构信息。然而，是否可以将蛋白质编码器整合到LLMs中以增强其对蛋白质的理解尚未被探索。为了弥合这一差距，我们提出了EvoLlama，一个多模态框架，连接了基于结构的编码器、基于序列的蛋白质编码器和LLM，用于蛋白质理解。EvoLlama由一个ProteinMPNN结构编码器、一个ESM-2蛋白质序列编码器、一个用于对齐蛋白质和文本表示的多模态投影器以及一个Llama-3文本解码器组成。为了训练EvoLlama，我们在面向蛋白质的指令和通过自然语言指令模板表述的蛋白质性质预测数据集上对其进行微调。我们的实验表明，EvoLlama的蛋白质理解能力得到了显著增强，在零样本设置下，平均比其它微调过的面向蛋白质的LLMs高出1%-8%，并平均超过具有监督微调的最先进基线6%。在蛋白质性质预测数据集上，我们的方法取得了与最先进的任务特定基线相竞争的有前景的结果。我们将在未来版本中发布我们的代码。

英文摘要：

Current Large Language Models (LLMs) for understanding proteins primarily treats amino acid sequences as a text modality. Meanwhile, Protein Language Models (PLMs), such as ESM-2, have learned massive sequential evolutionary knowledge from the universe of natural protein sequences. Furthermore, structure-based encoders like ProteinMPNN learn the structural information of proteins through Graph Neural Networks. However, whether the incorporation of protein encoders can enhance the protein understanding of LLMs has not been explored. To bridge this gap, we propose EvoLlama, a multimodal framework that connects a structure-based encoder, a sequence-based protein encoder and an LLM for protein understanding. EvoLlama consists of a ProteinMPNN structure encoder, an ESM-2 protein sequence encoder, a multimodal projector to align protein and text representations and a Llama-3 text decoder. To train EvoLlama, we fine-tune it on protein-oriented instructions and protein property prediction datasets verbalized via natural language instruction templates. Our experiments show that EvoLlama's protein understanding capabilities have been significantly enhanced, outperforming other fine-tuned protein-oriented LLMs in zero-shot settings by an average of 1%-8% and surpassing the state-of-the-art baseline with supervised fine-tuning by an average of 6%. On protein property prediction datasets, our approach achieves promising results that are competitive with state-of-the-art task-specific baselines. We will release our code in a future version.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。