OCG

OCG generates overlapping clusters in protein-protein interaction (PPI) networks to represent multifunctional proteins and capture their context-dependent interactions.

Key Features:

• Overlapping clustering: assigns proteins to multiple clusters to represent multifunctionality.
• Hierarchical ascending algorithm: decomposes networks using a hierarchical ascending algorithm.
• Initial overlapping classes: creates initial overlapping classes that cover the graph.
• Modularity-based fusion: iteratively fuses classes using an extension of Newman's modularity function adapted for overlaps.
• Fusion stopping criterion: stops merging when further fusion does not improve the modularity.
• Multifunctionality representation: captures multifunctional proteins at intersections of overlapping clusters.
• Validation and performance: validated on simulated graphs and real-world PPI networks, including human PPI networks, demonstrating superior performance in revealing multifunctional proteins.

Scientific Applications:

• Functional organization analysis: analyze the functional organization of biological systems by reflecting complex interactions and multifaceted protein roles.
• Multifunctional protein identification: identify and characterize multifunctional proteins within PPI networks.
• Network biology and protein function studies: support studies in protein function and network biology where overlapping functionalities are essential.

Methodology:

OCG decomposes a network into overlapping clusters by creating initial overlapping classes that cover the graph and iteratively fusing them with a hierarchical ascending algorithm using an extension of Newman's modularity function adapted for overlaps until modularity no longer increases; validation was performed on simulated graphs and real-world PPI networks.