pathwayPCA

pathwayPCA performs integrative pathway analysis of multi-omics datasets to estimate sample-specific pathway activities, select pathway-relevant genes, and test associations with continuous, binary, and survival outcomes.

Key Features:

• Supervised and adaptive elastic-net sparse PCA: Employs supervised and adaptive elastic-net sparse principal component analysis (PCA) for feature selection and extraction of pathway activity components.
• Multi-omics integration: Integrates multiple types of molecular data to derive pathway-level summaries across omics platforms.
• Outcome types supported: Tests associations between pathway activities and continuous, binary, and survival outcomes.
• Gene selection and sample-specific pathway activity estimation: Performs gene selection and provides sample-specific pathway activity scores.
• Sex-specific effects and prognostic modeling: Identifies sex-specific pathway effects (e.g., reported in kidney cancer) and supports integrative prognostic modeling.
• Performance benchmarking: Demonstrated superior performance in identifying disease-associated pathways in simulated and real datasets.
• Visualization: Provides functions for visualizing pathway activity patterns.

Scientific Applications:

• Integrative multi-omics analysis (CPTAC): Applied to integrative analyses of multi-omics datasets such as those from the Clinical Proteomic Tumor Analysis Consortium (CPTAC).
• Disease-associated pathway identification: Identifies pathways associated with disease phenotypes in simulated and empirical datasets.
• Biomarker discovery and prognostic modeling: Selects pathway-relevant genes and constructs integrative models for prognosis prediction.
• Sex-specific pathway analysis (kidney cancer): Detects sex-specific pathway effects as demonstrated in kidney cancer studies.
• Outcome association testing: Associates pathway activities with continuous, binary, and survival endpoints.

Methodology:

Implements supervised and adaptive elastic-net sparse principal component analysis (PCA) for dimensionality reduction and gene selection to derive sample-specific pathway activity scores from multiple molecular data types.