DriveWays

DriveWays identifies overlapping cancer driver modules by detecting overlapping gene modules using a seed-and-extend heuristic that integrates the IntAct protein-protein interaction network with mutual exclusivity, coverage, and network connectivity information to recover biologically relevant pathways.

Key Features:

• Formal problem definition (ODMIC): Defines the Overlapping Driver Module Identification in Cancer (ODMIC) problem and states it is NP-hard.
• Seed-and-extend heuristic: Uses a seed-and-extend algorithm that begins with seeds of known driver genes and extends them to identify larger overlapping modules.
• Integration of biological data: Incorporates the IntAct protein-protein interaction (PPI) network together with mutual exclusivity, coverage, and network connectivity information of genes.
• Performance validation: Evaluated on The Cancer Genome Atlas (TCGA) pan-cancer data and reported superior recovery of known cancer driver genes and stronger enrichment for reference pathways compared to existing methods.
• Support for overlapping modules: Allows modules to overlap to better recover functional pathways and capture genes acting as hubs across distinct gene sets.

Scientific Applications:

• Pan-cancer driver pathway discovery: Identification of overlapping driver modules across TCGA pan-cancer datasets to map cancer-related pathways.
• Recovery of known drivers and pathway enrichment: Recovery and enrichment analysis of established cancer driver genes and reference pathways.
• Modeling gene pleiotropy and hubs: Representation of genes that participate in multiple molecular pathways and act as network hubs.
• Mechanistic and translational studies: Use in studies aiming to elucidate disease mechanisms and to prioritize pathway-based therapeutic hypotheses.

Methodology:

Formulates the OMDIC problem (NP-hard) and applies a seed-and-extend heuristic starting from known driver gene seeds, integrating IntAct PPI data and metrics of mutual exclusivity, coverage, and network connectivity, with evaluation on TCGA pan-cancer data.