A sequence-based deep learning framework (PepInter) for protein–peptide interaction representation learning with pretrained protein language models

Protein-peptide Interactions play essential roles in diverse cellular processes, yet their systematic identification remains challenging due to the limited availability of experimentally annotated protein-peptide interaction data. To address this challenge, we present PepInter, a sequence-based Deep Learning framework that leverages large-scale pretraining on structurally derived pseudo protein-peptide pairs to learn interaction-relevant representations. Specifically, energy-dominant peptide fragments are extracted from protein complexes curated from non-redundant Protein Data Bank structures, enabling the construction of pseudo protein-peptide interaction pairs that capture interface interaction patterns shared with canonical protein-protein interactions. This strategy allows the model to acquire interaction-aware priors in the absence of large-scale annotated protein-peptide complex datasets. Built upon the ESM Cambrian architecture, PepInter adopts a two-stage pretraining strategy. Following pretraining, PepInter is fine-tuned for both protein-peptide interaction classification and peptide bioactivity regression tasks. Across multiple benchmark datasets, PepInter consistently outperforms state-of-the-art methods and demonstrates strong generalization in identifying biologically meaningful PpIs. Case studies further highlight its ability to recover known interaction patterns and predict novel protein-peptide interactions. Together, these results establish PepInter as a scalable and effective framework for protein-peptide interaction prediction, with strong potential to accelerate peptide-based drug discovery. Protein-peptide interactions are crucial for cellular functions yet identifying them systematically is hindered by scarce experimental data. Here, the authors introduce PepInter, a deep learning framework that uses pretraining on pseudo protein-peptide pairs to enhance interaction prediction.