associationSubgraphs

associationSubgraphs visualizes and identifies subgraph clusters within high-dimensional association matrices to reveal patterns of variable associations, including disease multimorbidity.

Key Features:

• Network percolation and clustering: Employs network percolation combined with clustering algorithms to detect connected components as association thresholds change.
• Dynamic cutoff exploration: Represents the entire clustering dynamics and provides subgraphs across all cutoff values simultaneously.
• Subgraph-focused detection: Identifies subsets of variables that exhibit stronger and more frequent internal associations than with variables outside the subset.
• Application to phenome-wide multimorbidity matrices: Applies to phenome-wide association matrices derived from electronic health records to analyze multimorbidity patterns.
• R package implementation: Implements the algorithmic and visualization components in an R package.

Scientific Applications:

• High-dimensional data analysis: Suited for analysis of large-scale datasets such as genomics, proteomics, and electronic health records.
• Network medicine: Facilitates discovery of disease comorbidity patterns from multimorbidity association matrices.
• Phenome-wide studies: Enables visualization and analysis of phenome-wide association data to investigate complex trait associations.

Methodology:

AssociationSubgraphs uses network percolation combined with clustering algorithms and dynamically adjusts cutoff values to reveal subgraphs representing clusters of variables with significant interconnections.