jNMF

jNMF performs joint matrix factorization to integrate multi-platform genomic profiles (DNA methylation (DM), gene expression (GE), and microRNA expression) into a shared coordinate system for discovery of multi-dimensional modules (md-modules) and biological associations.

Key Features:

• Joint Matrix Factorization Technique: Uses joint matrix factorization to project heterogeneous genomic datasets onto a common coordinate system.
• Projection onto a Shared Coordinate System: Identifies variables that are highly weighted in the same direction across platforms to define multi-dimensional modules (md-modules).
• Discovery of Combinatorial Patterns: Integrates multiple layers of genomic data to uncover combinatorial patterns and associations between different regulatory mechanisms.

Scientific Applications:

• Pathway Analysis: Reveals perturbed pathways that may be missed when analyzing single data types alone, aiding interpretation in complex diseases such as cancer.
• Functional Associations: Identifies significant correlations and likely functional associations among genomic variables within md-modules.
• Clinical Subgroup Discovery: Enables identification of clinically relevant patient subgroups by integrating multi-omic profiles.

Methodology:

Project multi-platform genomic profiles onto a shared coordinate system to identify md-modules of correlated variables; demonstrated by integrating DM, GE, and microRNA expression in 385 ovarian cancer samples from The Cancer Genome Atlas.