ProtLLM: An Interleaved Protein-Language LLM with Protein-as-Word Pre-Training

TL;DR

ProtLLM is a novel large language model that integrates protein and natural language understanding through a unique interleaved training approach, enabling it to perform well on protein tasks and zero-shot protein-language applications.

Contribution

It introduces a dynamic protein mounting mechanism and a protein-as-word modeling approach, along with a large-scale interleaved dataset for comprehensive protein and text learning.

Findings

Achieves superior performance on protein-centric tasks

Demonstrates zero-shot and in-context learning for protein-language tasks

Outperforms protein-specialized baselines

Abstract

We propose ProtLLM, a versatile cross-modal large language model (LLM) for both protein-centric and protein-language tasks. ProtLLM features a unique dynamic protein mounting mechanism, enabling it to handle complex inputs where the natural language text is interspersed with an arbitrary number of proteins. Besides, we propose the protein-as-word language modeling approach to train ProtLLM. By developing a specialized protein vocabulary, we equip the model with the capability to predict not just natural language but also proteins from a vast pool of candidates. Additionally, we construct a large-scale interleaved protein-text dataset, named InterPT, for pre-training. This dataset comprehensively encompasses both (1) structured data sources like protein annotations and (2) unstructured data sources like biological research papers, thereby endowing ProtLLM with crucial knowledge for understanding proteins. We evaluate ProtLLM on classic supervised protein-centric tasks and explore its novel protein-language applications. Experimental results demonstrate that ProtLLM not only achieves superior performance against protein-specialized baselines on protein-centric tasks but also induces zero-shot and in-context learning capabilities on protein-language tasks.