Knowledge-Augmented Spectral Hypergraph Learning for Protein Complex Identification in Network-Based Drug Discovery.

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• Protein complexes are crucial for cellular functions and drug discovery.
• Existing computational methods struggle with complex cooperation patterns and biological constraints in protein-protein interaction (PPI) networks.

Purpose of the Study:

• To introduce KSHL-PC, a knowledge-augmented spectral hypergraph learning framework for accurate protein complex identification.
• To overcome limitations of pairwise interaction and heuristic clustering approaches.

Main Methods:

• Constructing clique-induced hypergraphs to model multi-protein cooperation.
• Employing spectral hypergraph embedding for multi-scale structural dependency capture.
• Integrating Gene Ontology annotations and temporal activity signals via knowledge-enhanced feature transformation.

Main Results:

• KSHL-PC significantly outperforms state-of-the-art methods in F1-score, AUPRC, and complex-level accuracy.
• Predicted protein complexes demonstrate strong functional coherence, subcellular colocalization, and transcriptional coordination.
• The framework enables end-to-end complex scoring without manual rules.

Conclusions:

• KSHL-PC offers an effective approach for identifying biologically relevant protein complexes.
• The method advances network-based drug target discovery by improving protein complex identification.
• Publicly available code and datasets facilitate further research and application.