Pathway extraction

Pathway extraction extracts relevant subnetworks from metabolic networks to predict and analyze biological pathways by connecting input entities such as compounds, reactions, EC numbers, and enzyme-coding genes as a component of the Network Analysis Tools (NeAT) suite.

Key Features:

• Subgraph Extraction: Extracts relevant subnetworks from complex metabolic networks and accepts data types including gene expression, protein levels, operons, and phylogenetic profiles.
• Hybrid Strategy for Pathway Recovery: Combines random walk-based graph reduction with shortest paths algorithms and recovered known metabolic pathways with approximately 77% accuracy on Saccharomyces cerevisiae using MetaCyc data.
• Handling of Pool Metabolites: Mitigates shortcuts caused by pool metabolites (e.g., water and cofactors) by penalizing highly connected compounds and distinguishing main versus side reactants based on chemical structure.
• Integration with KEGG RPAIR Data: Incorporates KEGG RPAIR reactant-pair categorizations (e.g., main, trans, cofac) to inform pathfinding, improving pathway recovery rates up to 93% for Escherichia coli pathways.
• Weighted Pathfinding Approaches: Assigns weights to compounds based on connectivity to improve relevance and enable correct inference of longer pathways with multiple intermediate reactions.
• Multiple Pathway Calculation Options: Computes shortest paths and k-shortest paths between specified seed nodes such as compounds or reactions.
• Customizable Network Input: Operates on KEGG LIGAND, KEGG RPAIR, and MetaCyc networks and accepts custom metabolic network inputs.

Scientific Applications:

• Metabolic Pathway Reconstruction: Reconstructs metabolic pathways across organisms using pathway databases and genome-derived inputs such as EC numbers and enzyme-coding genes.
• Systems Biology and Network Analysis: Supports systems biology and computational biology studies that require subnetwork extraction and pathway prediction.
• Drug Discovery and Disease Modeling: Aids drug discovery and disease modeling by identifying biochemical routes and potential enzymatic intervention points.
• Synthetic Biology: Enables inference of biosynthetic routes and identification of targets for pathway engineering in synthetic biology applications.

Methodology:

Combines random walk-based graph reduction with shortest-paths algorithms (including k-shortest paths), uses weighted graphs that penalize highly connected compounds, differentiates main and side reactants based on chemical structure, and integrates KEGG RPAIR reactant-pair categorizations (main, trans, cofac) to inform pathfinding.