FactorViz

FactorViz visualizes and reports results from methylome decomposition experiments to interpret latent methylation components (LMCs) recovered from reference-free deconvolution of bulk DNA methylation data.

Key Features:

• Visualization and reporting: Generates visualizations and reports of methylome decomposition results derived from bulk DNA methylation data.
• Deconvolution compatibility: Supports results produced by reference-free deconvolution methods MeDeCom, RefFreeCellMix, and EDec.
• Latent methylation component interpretation: Interprets and annotates LMCs that are specific to different cell types.
• Confounder adjustment: Performs confounder adjustment using independent component analysis (ICA).
• Feature selection: Performs feature selection using the DecompPipeline.
• Parameter exploration: Supports analysis of decompositions obtained with multiple parameter settings to identify robust LMCs.
• Clinical association analysis: Associates methylation components with clinical parameters.
• Cell proportion inference: Infers proportions of stromal cells and tumor-infiltrating immune cells from LMCs.

Scientific Applications:

• Dissection of tumor heterogeneity: Recovery of cell-type-specific methylation signals and estimation of cell proportions in tumors.
• Analysis of complex tissues and mixtures: Interpretation of bulk DNA methylation data when purified cell-type profiles are not available.
• Clinical association studies: Linking latent methylation components to clinical metadata and parameters.
• TCGA lung cancer methylome analysis: Identification of stromal and tumor-infiltrating immune cell proportions in lung cancer methylomes from The Cancer Genome Atlas (TCGA).

Methodology:

Data preprocessing with confounder adjustment using independent component analysis (ICA) and feature selection via DecompPipeline; deconvolution using MeDeCom, RefFreeCellMix, or EDec with multiple parameter settings to identify latent methylation components (LMCs); downstream association analyses to link components to clinical parameters and infer cell-type proportions.