DEXOM

DEXOM reconstructs and enumerates diverse context-specific genome-scale metabolic network submodels by integrating multi-omics data to capture alternative metabolic states.

Key Features:

• MATLAB implementation: Provided as a MATLAB library for computational reconstruction and enumeration of metabolic networks.
• Multi-omics integration: Integrates experimental multi-omics data into genome-scale metabolic networks (GSMNs) to inform model reconstruction.
• Context-specific sub-network extraction: Extracts sub-networks from generic GSMNs that are most consistent with available experimental data while enforcing biochemical constraints.
• Formalized enumeration problem: Formalizes the multiplicity of optimal context-specific networks that can fit experimental data.
• Diversity-based enumeration strategy: Implements a unified strategy to enumerate diverse optimal networks and explore alternative metabolic explanations.
• Multiple enumeration strategies: Employs various computational strategies for network enumeration to capture solution diversity.
• Validation on data: Evaluated using both simulated and real datasets to assess biological relevance.
• Ensemble modeling: Supports use of ensembles of metabolic networks to improve predictive tasks such as essential gene prediction.
• COBRA compatibility: Compatible with COBRA Toolbox 3.0 for integration with established constraint-based modeling workflows.

Scientific Applications:

• Context-specific metabolic reconstruction: Reconstructs tissue-, cell-type-, or condition-specific metabolic models to reflect active reactions under specific biological contexts.
• Exploration of alternative metabolic states: Enumerates multiple optimal networks to reveal alternative metabolic explanations for the same data.
• Essential gene prediction in Saccharomyces cerevisiae: Uses ensembles of metabolic networks to improve in silico predictions of essential genes in Saccharomyces cerevisiae.
• Pathway analysis in human cancer cell lines: Detects alternative enriched pathways across different human cancer cell lines by comparing enumerated networks.
• Addressing limitations of expression data: Accounts for discrepancies between gene expression and metabolic activity arising from post-transcriptional modifications and protein degradation rates.

Methodology:

Implemented in MATLAB, DEXOM integrates multi-omics into GSMNs, extracts context-specific sub-networks consistent with experimental data under biochemical constraints, formalizes and enumerates multiple optimal networks using diversity-based enumeration strategies and various enumeration methods, and has been tested on simulated and real datasets.