ROBUST

ROBUST identifies candidate disease-module subnetworks in molecular interaction networks, such as protein-protein interaction (PPI) networks, by enumerating diverse prize-collecting Steiner trees (PCSTs) to improve robustness and functional relevance of disease module mining methods (DMMMs).

Key Features:

• Disease module extraction: Extracts subgraphs that serve as candidate disease mechanisms from molecular interaction networks such as PPI networks.
• Robustness: Consistently produces stable and reliable disease modules across multiple runs on identical datasets.
• PCST enumeration: Leverages enumeration of diverse prize-collecting Steiner trees (PCSTs) to identify subnetworks that optimally connect designated "prize" nodes while minimizing total edge cost.
• Scalability: Demonstrates scalability suitable for large-scale applications in complex biological networks.
• Functional validation: Modules show high KEGG gene set enrichment scores and significant overlap with disease genes in DisGeNET.

Scientific Applications:

• Candidate mechanism discovery: Identification of candidate disease mechanisms and subnetworks from molecular interaction data.
• Robustness benchmarking: Assessment and improvement of robustness in disease module mining methods (DMMMs).
• Pathway and disease-gene interpretation: Functional interpretation of modules via KEGG pathway enrichment and overlap with DisGeNET disease genes.
• Large-scale network analysis: Application to large, complex protein-protein interaction networks for disease-module discovery.

Methodology:

Enumerates diverse prize-collecting Steiner trees (PCSTs) to extract subnetworks that optimally connect specified "prize" nodes while minimizing total edge cost.