UNCOVER

UNCOVER identifies sets of somatic genetic alterations that exhibit mutually exclusive complementary functional associations and associates those gene sets with quantitative target profiles (e.g., genetic perturbations or clinical phenotypes) in cancer.

Key Features:

• Mutual Exclusivity Analysis: UNCOVER employs mutual exclusivity to detect gene sets whose alterations occur in a complementary manner across tumors, indicating shared functional roles.
• Quantitative Target Profiles: The tool integrates quantitative target profiles derived from genetic perturbations or clinical phenotypes to associate gene sets with functional outcomes.
• Combinatorial Formulation: UNCOVER formulates the identification of mutually exclusive alteration sets associated with a quantitative target as a combinatorial optimization problem.
• Two Algorithms: It implements two algorithms developed to solve the combinatorial problem efficiently and to identify high-quality solutions.
• Experimental Validation: The method has been evaluated across scenarios and shown to find alteration sets significantly associated with functional targets using statistical evaluation criteria.
• Scalability and Large-Scale Datasets: UNCOVER can process large datasets comprising thousands of target profiles from cancer cell lines, including data from the Achilles project and the Genomics of Drug Sensitivity in Cancer project.

Scientific Applications:

• Identification of complementary functional gene sets: Detects gene sets with mutually exclusive alterations that suggest pathway-level functional associations in tumors.
• Correlation with drug sensitivity and perturbations: Associates somatic alteration patterns with drug sensitivity and genetic perturbation profiles to support analysis of datasets such as Achilles and GDSC.
• Prioritization of therapeutic targets and mechanistic studies: Supports discovery and prioritization of candidate therapeutic targets and mechanistic insights by linking alteration patterns to quantitative phenotypes.

Methodology:

UNCOVER employs mutual exclusivity analysis, integrates quantitative target profiles (genetic perturbations or clinical phenotypes), formulates the problem as a combinatorial optimization task, and implements two algorithms to solve it.