Large language models (LLMs) have significantly advanced protein representation learning. However, their capacity to interpret and design antibodies through natural language remains limited. To address this challenge, we present AFD-Instruction, the first large-scale instruction dataset with functional annotations tailored to antibodies. This dataset encompasses two key components: antibody understanding, which infers functional attributes directly from sequences, and antibody design, which enables de novo sequence generation under functional constraints. These components provide explicit sequence-function alignment and support antibody design guided by natural language instructions. Extensive instruction-tuning experiments on general-purpose LLMs demonstrate that AFD-Instruction consistently improves performance across diverse antibody-related tasks. By linking antibody sequences with textual descriptions of function, AFD-Instruction establishes a new foundation for advancing antibody modeling and accelerating therapeutic discovery.
@article{arxiv.2602.04916,
title = {AFD-INSTRUCTION: A Comprehensive Antibody Instruction Dataset with Functional Annotations for LLM-Based Understanding and Design},
author = {Ling Luo and Wenbin Jiang and Hongyuan Chang and Xinkang Wang and Xushi Zhang and Yueting Xiong and Mengsha Tong and Rongshan Yu},
journal= {arXiv preprint arXiv:2602.04916},
year = {2026}
}