Prodigy

Prodigy prioritizes patient-specific cancer driver genes by quantifying the impact of mutated genes on deregulated pathways using mutation, expression, pathway, and protein–protein interaction data to support personalized cancer therapy.

Key Features:

• Implementation: Implemented as an R package for computational analysis of patient-level cancer genomics.
• Patient-Specific Driver Gene Ranking: Ranks mutated genes based on their impact on deregulated pathways within individual tumors.
• Integration of Multi-Omics Data: Integrates expression and mutation profiles with known biological pathways and protein-protein interactions.
• Prize-Collecting Steiner Tree Model: Uses the prize-collecting Steiner tree model to quantify the influence of mutated genes on deregulated pathways.
• Aggregation and Ranking: Aggregates pathway-level impact scores to produce an overall ranking of candidate driver genes per patient.
• Performance and Validation: Validated across five TCGA cohorts (The Cancer Genome Atlas) involving over 2500 patients, showing better performance than existing methods and network centrality measures and identifying both common and rare driver genes.
• Pleiotropic Effect Identification: Pinpoints pleiotropic effects of driver genes to inform disease mechanisms and potential therapeutic targets.

Scientific Applications:

• Driver Gene Identification: Improves identification of driver genes that contribute to tumor development at the individual-patient level.
• Pathway Dissection: Provides insights into the molecular pathways disrupted by patient-specific mutations.
• Target Discovery and Drug Development: Prioritizes candidate therapeutic targets to guide targeted therapy selection and drug discovery efforts.
• Personalized Treatment Planning: Supports personalized treatment strategies by ranking actionable or influential mutated genes per patient.

Methodology:

Ranks mutated genes by measuring their impact on deregulated pathways using expression and mutation profiles integrated with known pathways and protein-protein interaction networks, applies the prize-collecting Steiner tree model to quantify pathway impacts and aggregates these impacts into per-patient gene rankings, and was evaluated across five TCGA cohorts (over 2500 patients) against existing methods and network centrality measures.