NMFNA

NMFNA applies non-negative matrix factorization with a graph-regularized constraint to integrate methylation (ME) and copy number variation (CNV) networks constructed from TCGA using Pearson Correlation Coefficient, detect core communities and characteristic genes in pancreatic cancer, and perform gene ontology and pathway enrichment analyses.

Key Features:

• Graph Regularized Constraint: Incorporates a graph-regularized constraint into the NMF framework to preserve network topology during factorization and improve community detection.
• Network Construction: Builds three pancreatic cancer networks—Methylation Network (ME), Copy Number Variation Network (CNV), and a bipartite network connecting ME and CNV—using Pearson Correlation Coefficient on TCGA data.
• Community Detection: Detects core communities within the ME, CNV, and bipartite networks by leveraging the graph-regularized NMF formulation.
• Gene Identification: Identifies characteristic genes associated with detected communities in pancreatic cancer for downstream analysis.
• Enrichment Analysis: Performs gene ontology enrichment analysis and pathway enrichment analysis on community-associated genes to interpret biological functions and pathways.

Scientific Applications:

• Biomarker Discovery: Identifying novel biomarkers for pancreatic cancer by extracting characteristic genes from network communities.
• Mechanistic Insight: Understanding gene-environment and multi-omic interactions in pancreatic cancer by integrating methylation and CNV data.
• Therapeutic Target Exploration: Exploring potential therapeutic targets by linking community-identified genes to enriched pathways and functions.

Methodology:

Construct networks from TCGA methylation and CNV data using Pearson Correlation Coefficient, apply graph-regularized non-negative matrix factorization to detect communities and characteristic genes, and perform gene ontology and pathway enrichment analyses.