Protein-Protein Interaction (PPI) prediction is a key task in uncovering cellular functional networks and disease mechanisms.
A novel supervised contrastive multimodal framework, SCMPPI, is proposed for PPI prediction. It integrates protein sequence features and PPI network topology information, enhancing prediction performance.
SCMPPI outperforms existing state-of-the-art methods in accuracy and AUC metrics, and demonstrates strong generalization in cross-species prediction.
SCMPPI has been successfully applied to CD9 networks, the Wnt pathway, and cancer-specific networks, providing a reliable tool for disease target discovery.