ELF-DPC

ELF-DPC identifies protein complexes in protein-protein interaction (PPI) networks by integrating ensemble learning, core mining, and graph-based extension to improve detection accuracy and biological relevance.

Key Features:

• Weighted PPI Network Construction: Constructs a weighted PPI network by integrating topological and biological information.
• Integration of Unsupervised and Supervised Learning: Combines unsupervised and supervised learning approaches to capture diverse signals for complex detection.
• Protein Complex Core Mining: Identifies potential cores of protein complexes using a tailored core mining strategy.
• Ensemble Learning Model Integration: Employs an ensemble model that combines structural modularity measures with a trained voting regressor.
• Graph Heuristic Search Strategy: Extends identified cores into full protein complexes using a graph heuristic search strategy.

Scientific Applications:

• Cellular Organization Analysis: Detects biologically meaningful protein complexes to support studies of cellular organization and processes.
• Detection across Topologies: Identifies protein complexes with varying topological structures within PPI networks.
• Benchmarking: Demonstrated to outperform twelve state-of-the-art protein complex detection approaches in experimental evaluations.
• Functional Validation: Supports functional enrichment analysis that confirms the biological significance of detected complexes.

Methodology:

Computational steps explicitly include weighted PPI network construction combining topological and biological information; protein complex core mining; integration of an ensemble model combining structural modularity with a trained voting regressor; and extension of cores into complete complexes via graph heuristic search.