SeekRBP: Leveraging Sequence-Structure Integration with Reinforcement Learning for Receptor-Binding Protein Identification.

SeekRBP identifies receptor-binding proteins in Vibrio phages by integrating sequence and structure, using dynamic hard-negative sampling to improve host prediction.

• Why it matters: RBP discovery is hindered by extreme sequence divergence that breaks homology-based methods, and by class imbalance plus hard negatives from heterogeneous tail proteins.
• What they did: SeekRBP treats negative sampling as sequential decision-making, using a multi-armed bandit to prioritize informative non-RBPs during training, with multimodal fusion of protein language and structural embeddings.
• The result: Benchmarking shows SeekRBP consistently outperforms static sampling strategies, and a Vibrio phage case study indicates it can identify RBPs that improve host prediction.

AI summary of the abstract · may contain errors · not medical advice · read the paper.