PPS

PPS computes the pair-path subscore to quantify how specific paths in Gaussian graphical models contribute to observed Pearson correlations in biological network data from cross-sectional studies.

Key Features:

• Pair-Path Subscore (PPS): Computes a metric that quantifies the significance of specific paths between nodes in a Gaussian graphical model and the contribution of individual paths to the Pearson correlation between their terminal nodes.
• Network-Level Interpretation: Evaluates contributions of entire paths within the network topology rather than focusing solely on pairwise dependencies.
• Application to Metabolomics and HAPO: Applies PPS to human metabolomics datasets, including the Hyperglycemia and adverse pregnancy outcome (HAPO) study, to validate known biological relationships.
• Exploratory Analysis: Identifies significant paths within networks to enable generation of novel biological hypotheses.
• Implementation in R: Provided as the pps_r R package for computation of PPS on Gaussian graphical models.

Scientific Applications:

• Metabolomics Research: Uses PPS on metabolomic data to reveal relationships among metabolites and inform metabolic pathway interpretation.
• Network Biology: Investigates how specific paths within biological networks contribute to observed correlations and network dynamics.
• Hypothesis Generation: Facilitates formulation of new hypotheses about node relationships and subnetworks in systems biology studies.

Methodology:

Calculation of the pair-path subscore for paths within a Gaussian graphical model to assess the relative importance of these paths in determining Pearson correlations between terminal nodes, with validation using human metabolomics data from the HAPO study.