MADGiC

MADGiC applies an empirical Bayesian hierarchical model to somatic mutation data to prioritize driver genes in cancer by deriving posterior probabilities for each gene.

Key Features:

• Integration of Frequency and Functional Impact: Integrates mutation frequency and functional impact within genes to prioritize mutations more likely to be drivers.
• Spatial Pattern Analysis: Analyzes the spatial distribution of mutations within a gene to distinguish patterns associated with functional consequences.
• Enhanced Background Model: Implements a refined background model that accommodates multiple factors to improve power and precision over simple frequency-based models.
• Implementation: Provided as an R package that implements the empirical Bayesian hierarchical model and computes posterior probabilities per gene.

Scientific Applications:

• Driver gene discovery in cancer cohorts: Prioritizes somatic mutations to identify candidate driver genes in cancer genomics studies.
• TCGA dataset analysis: Has been applied to The Cancer Genome Atlas (TCGA) datasets, including ovarian and lung cancer, to identify candidate drivers.
• Therapeutic target prioritization: Supports selection of genes for downstream functional studies and potential therapeutic targeting.

Methodology:

Fits an empirical Bayesian hierarchical model to mutation data to derive posterior probabilities per gene, integrating mutation frequency, functional impact, spatial distribution within genes, and a refined background model; performance evaluated using simulation studies.