CCPROMISE

CCPROMISE integrates genomic, epigenomic, and transcriptomic data using canonical correlation analysis and the PROMISE framework to associate molecular measurements with multiple clinical endpoints and identify genes with biologically significant associations.

Key Features:

• Canonical Correlation Analysis (CCA): Computes canonical correlation scores that quantify associations between two different molecular data types.
• Projection onto the Most Interesting Evidence (PROMISE): Applies PROMISE to assess the statistical evidence that canonical correlation patterns relate to multiple clinical endpoints.
• Multi-omics Integration: Integrates genomic, epigenomic, and transcriptomic datasets to enable cross-platform association analyses with clinical endpoints.
• Statistical Rigor: Controls type I error rates near nominal levels across null settings where no molecular-endpoint association exists.
• High Statistical Power: Demonstrates superior power versus competing methods in 99% of alternative scenarios, with a greater-than-30% power advantage in over one-third of cases.

Scientific Applications:

• Simulation Studies: Validated by simulation studies to assess type I error control and statistical power.
• Pediatric Leukemia and Biomarker Discovery: Applied to pediatric leukemia data to identify genes whose molecular profiles associate with multiple related clinical endpoints, informing potential biomarker or therapeutic target discovery.

Methodology:

Computes canonical correlation scores between molecular datasets and then applies PROMISE to evaluate the statistical significance of these scores with respect to multiple clinical endpoints.