cowmunity1.0

Cowmunity Model: Community Metabolic Modeling with Molecular Docking Integration

Overview

What We've Done

1. Community Metabolic Modeling: Built a bilevel optimization model that simulates three rumen bacteria:
   • Methanobrevibacter gottschalkii (MGK): Methanogen that produces methane
   • Prevotella ruminicola (PRM): Fibrolytic bacterium
   • Ruminococcus flavefaciens (RFL): Fibrolytic bacterium
2. Molecular Docking Integration: Integrated structure-based molecular docking predictions directly into the metabolic model by:
   • Converting binding affinities (Kd) to inhibition factors
   • Accounting for drug-enzyme binding saturation
   • Applying relaxation factors to bridge in silico predictions and in vivo reality
   • Mapping docking results to metabolic reactions
3. Combined Approach: Simultaneously applies:
   • Direct enzyme effects from docking (structure-based predictions)
   • Indirect community effects from literature (community interactions, substrate availability)
4. Metabolite Exchange: Models inter-species metabolite sharing (H₂, CO₂, formate, amino acids, etc.) to create a true community model rather than three separate models.

Installation

Prerequisites

• Python 3.7 or higher
• GAMSpy Academic License (free for academic use)

Step 1: Install Python Packages

Step 2: Register for GAMSpy License

1. Register for a free academic license at: https://academic.gams.com/
2. Install your license:

Running the Model

Basic Usage

1. Navigate to the project directory:

1. Run the main script:

1. Select a treatment option when prompted:
   • 0: No treatment (baseline)
   • 1: Imidazole
   • 2: L-Carnitine
   • 3: Methyl Jasmonate
   • 4: Propylpyrazine

What Happens When You Run

1. Load and process the three SBML metabolic models
2. Set up community exchange constraints
3. Apply treatment-specific constraints (docking-based + literature-based)
4. Solve the bilevel optimization problem
5. Save results to results/ directory
6. Display key outputs (biomass, methane flux)

Output

• mgk_records.csv - All reaction fluxes for M. Gottschalkii
• prm_records.csv - All reaction fluxes for P. ruminicola
• rfl_records.csv - All reaction fluxes for R. flavefaciens

Project Structure

CowmunityModel/
├── main.py                          # Main entry point
├── Cowmunity.py                     # Core model implementation
├── cow.txt                          # ASCII art (optional)
├── model files/                     # SBML metabolic models
│   ├── M. gottschalkii.xml
│   ├── P. ruminicola.xml
│   └── R. flavefaciens.xml
├── docking_integration/              # Docking integration module
│   ├── apply_docking_constraints.py
│   ├── parse_docking_results.py
│   └── enzyme_reaction_mapper.py
└── docking_data/                     # Docking results
    └── cleaned_docking_results.csv

Key Features

• Bilevel Optimization: Outer problem maximizes community biomass; inner problems optimize individual species
• Structure-Based Predictions: Uses molecular docking data to predict enzyme inhibition
• Community Interactions: Models metabolite exchange between species
• Combined Constraints: Integrates docking predictions with literature-based effects
• Flexible Treatments: Test multiple small molecule treatments

Dependencies

• pandas - Data manipulation
• python-libsbml - SBML model parsing
• gamspy - Optimization framework (requires academic license)

litreture model

• To access litreture based model, simply use the model with litreture name on it.

Model Units

• Fluxes: mmol/gDCW·hr (millimoles per gram dry cell weight per hour)
• Biomass: gDCW/gDCW·hr (grams dry cell weight per gram dry cell weight per hour)
• Methane: mmol/gDCW·hr (can be converted to ml/gDCW·hr)

Citation

Authors

Troubleshooting

GAMSpy License Issues

• Ensure you have registered for an academic license
• Check that your license is properly installed: gamspy license status

Missing Files

• Ensure all SBML model files are in the model files/ directory
• Check that docking_data/cleaned_docking_results.csv exists for docking-based treatments

Solver Issues

• Default solver is IPOPT. If issues occur, ensure IPOPT is properly installed with GAMSpy
• Model solve time is limited to 20 seconds by default (can be modified in Cowmunity.py)

Contact