MEXCOwalk: mutual exclusion and coverage based random walk to identify cancer modules

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Genomic analyses of large cancer cohorts reveal significant mutational heterogeneity, complicating the identification of driver genes based solely on mutation profiles.
• Genes function within interconnected modules, suggesting that integrating functional connectivity can improve driver gene identification.
• Protein-protein interaction (PPI) networks offer valuable connectivity information, which, combined with mutation data, can enhance the accuracy of identifying cancer driver modules.

Purpose of the Study:

• To develop a novel computational method for identifying cancer driver modules by integrating diverse biological data.
• To improve the accuracy of cancer driver module identification beyond traditional mutation profile analysis.
• To leverage protein-protein interaction networks, mutation frequencies, and mutual exclusivity patterns for enhanced driver module discovery.

Main Methods:

• Developed an edge-weighted random walk-based approach named MEXCOwalk.
• Incorporated protein-protein interaction (PPI) network connectivity, gene mutation frequencies, mutual exclusivity, and coverage.
• Applied the method to TCGA (The Cancer Genome Atlas) pan-cancer data for module identification and validation.

Main Results:

• MEXCOwalk outperformed several state-of-the-art methods in identifying known cancer genes within modules.
• Generated modules demonstrated capability in classifying normal versus tumor samples.
• Identified modules were enriched for mutations specific to particular cancer types and could stratify patients into distinct risk groups.
• Discovered modules containing both established and putative cancer genes, including rarely mutated ones.

Conclusions:

• The MEXCOwalk approach effectively identifies cancer driver modules by integrating PPI networks and mutation data.
• This method enhances the accuracy of driver module discovery and provides clinically relevant patient stratification.
• MEXCOwalk offers a valuable tool for uncovering complex cancer gene interactions and identifying novel cancer drivers.