PIGNON

PIGNON performs functional enrichment analysis of quantitative proteomics by leveraging expression-weighted protein-protein interaction (PPI) networks and Gene Ontology (GO) annotations to detect differentially expressed, clustered functional terms.

Key Features:

• Graph-theory clustering: Uses a graph theory-based approach to identify clusters of proteins associated with specific Gene Ontology (GO) annotations within a PPI network.
• Expression-weighted PPI network: Maps protein differential expression levels onto an expression-weighted protein-protein interaction (PPI) network to incorporate quantitative proteomics data.
• Cluster-centric detection: Detects functional annotations where proteins are collectively clustered and differentially expressed even when individual proteins do not meet conventional significance thresholds.
• Statistical assessment: Assesses clustering significance using Monte Carlo sampling and approximates a normal distribution to estimate clustering significance.
• Multiple testing correction: Controls for multiple hypothesis testing by evaluating the false discovery rate (FDR) against a null model across thresholds.
• Complementary enrichment: Complements standard functional enrichment analyses by identifying GO terms missed by conventional per-protein significance methods.

Scientific Applications:

• Quantitative proteomics enrichment: Functional enrichment analysis of quantitative proteomics datasets to identify dysregulated biological processes.
• Breast cancer subtype analysis: Comparative analysis of molecular subtypes of breast cancer, including HER2+, triple-negative, and hormone receptor-positive types, to identify GO terms that are clustered and differentially expressed across subtypes.
• Dysregulated process discovery: Identification of dysregulated functions and processes in biological samples under varying experimental conditions.

Methodology:

Maps protein differential expression onto the PPI network, measures clustering of proteins that share GO annotations in an expression-weighted PPI network, uses Monte Carlo sampling with a normal approximation to estimate clustering significance, and evaluates false discovery rate against a null model.