iSFun

iSFun performs integrative dimension reduction analysis of high-dimensional omics data to identify shared and study-specific latent components across multiple independent studies.

Key Features:

• Integrative Analysis Techniques: Implements integrative sparse Principal Component Analysis (PCA), Partial Least Squares (PLS), and Canonical Correlation Analysis (CCA) for joint dimension reduction across datasets.
• Homogeneity and Heterogeneity Models: Supports both homogeneity and heterogeneity model formulations to represent shared versus study-specific signal structures.
• Contrasted Penalties: Incorporates magnitude- and sign-based contrasted penalties to impose sparsity and directional constraints on components.
• Meta-analysis and Stacked Analysis: Provides functionality for meta-analysis and stacked analysis of multiple compatible-study datasets.

Scientific Applications:

• Multi-study omics integration: Enables integrative analysis of high-dimensional omics data to reveal patterns and relationships obscured in single-study analyses.
• Discovery of shared and study-specific signals: Facilitates identification of shared biological signals and study-specific variation across independent datasets.

Methodology:

Implements integrative sparse PCA, Partial Least Squares (PLS), and Canonical Correlation Analysis (CCA) with magnitude- and sign-based contrasted penalties and supports homogeneity/heterogeneity models, meta-analysis, and stacked analysis.