IMCDriver

IMCDriver identifies and prioritizes personalized candidate driver genes in cancer by integrating known driver genes as priors with multi-omics data, including somatic mutations, gene expression profiles, and protein-protein interactions, to assess functional similarity between patients and genes.

Key Features:

• Integration of prior driver genes: Incorporates well-established driver genes as prior information to guide gene prioritization.
• Multi-omics similarity computation: Computes similarities between patients and genes using somatic mutations, gene expression profiles, and protein-protein interactions.
• Personalized gene prioritization: Prioritizes mutated genes for individual patients based on functional similarity to known driver genes.
• Inductive Matrix Completion (IMC) ranking: Applies Inductive Matrix Completion (IMC) to rank genes by their likelihood of being drivers in individual cancer cases.
• Detection of rare drivers: Identifies and prioritizes infrequently mutated or rare driver genes across patient populations.
• Benchmarking on consortium datasets: Demonstrated improved identification of cohort-level and patient-specific driver genes compared with existing methods on five diverse cancer datasets from the Cancer Genome Consortium.
• Novel driver discovery: Uncovered novel candidate driver genes implicated in cancer development.

Scientific Applications:

• Personalized oncology: Identification and ranking of candidate driver genes at the individual patient level to inform personalized analyses.
• Cohort-level discovery: Discovery and prioritization of driver genes across cancer cohorts for comparative studies.
• Rare mutation prioritization: Prioritization of rare or infrequently mutated driver genes that may be missed by frequency-based approaches.
• Functional interpretation: Linking somatic mutations and gene expression via protein-protein interactions to provide functional similarity-based interpretation of candidate drivers.

Methodology:

Integrates known driver genes as prior information; computes patient–gene similarities from somatic mutations, gene expression profiles, and protein-protein interactions; applies Inductive Matrix Completion (IMC) to rank genes by driver likelihood; evaluated on five cancer datasets from the Cancer Genome Consortium.