IUSMMT

IUSMMT performs mediation analysis to test whether multiple DNA methylation CpG sites influence cancer survival via a single gene's expression using mixed models and an intersection-union testing framework.

Key Features:

• R implementation: Implemented as an R-based procedure for statistical analysis.
• Multiple-exposure handling: Tests scenarios with multiple DNA methylation loci (CpG sites) as exposures and one gene expression as the mediator.
• Survival outcome modeling: Targets survival time as the outcome in mediation analyses.
• Mixed models: Employs mixed models to account for relevant random and fixed effects in the data.
• Intersection-Union testing: Uses an intersection-union testing approach implicit in the method name to combine component tests.
• Empirical three-component mixture null distribution: Adopts an empirical three-component mixture null distribution to determine mediation significance.
• Simulation-based evaluation: Performance and power were evaluated and compared to existing methods via extensive simulation studies.
• Application to TCGA: Applied to The Cancer Genome Atlas (TCGA) data across ten cancer types to identify mediating genes.

Scientific Applications:

• Mediator discovery: Identify genes whose expression mediates the effect of DNA methylation CpG sites on cancer survival.
• Cancer-type specificity: Detect cancer-type-specific mediators and regions exhibiting full mediation from CpG sites to survival risk.
• Prognostic prioritization: Prioritize candidate mediating genes for further study of prognosis and potential therapeutic targeting.
• Method benchmarking: Compare mediation-test power and effectiveness against existing methods via simulation studies.

Methodology:

R-based implementation using mixed models within an intersection-union testing framework and an empirical three-component mixture null distribution, evaluated by simulation studies and applied to TCGA datasets across ten cancer types.