Prediction of GO Terms Based on Partitioning PPI Networks into Highly Connected Components

Area of Science:

• Bioinformatics
• Computational Biology
• Network Analysis

Background:

• Accurate Gene Ontology (GO) term prediction is crucial for understanding protein function.
• Biological networks, such as protein-protein interaction (PPI) networks, are complex and require effective partitioning for analysis.

Purpose of the Study:

• To propose a novel two-phase method for GO term prediction.
• To integrate Variable Neighborhood Search (VNS) for biological network partitioning with functional annotation.

Main Methods:

• Phase 1: Employing a VNS metaheuristic to detect highly connected components in PPI networks, creating cohesive subnetworks.
• Phase 2: Utilizing network partitions for functional enrichment analysis with the DINGO tool for GO term prediction.

Main Results:

• The proposed approach effectively partitions biological networks, yielding high-quality subnetworks.
• Experimental results demonstrate the potential of the method for accurate GO term annotation.

Conclusions:

• The novel two-phase method enhances GO term prediction accuracy through effective network partitioning.
• The VNS framework is versatile and applicable to various network analysis tasks beyond biological networks.