This repository contains the software developed as part of the master’s thesis
“Classification of Gene Regulatory Networks Driving Shape Homeostasis”
by M.Sc. J. Esteban Pérez-Hidalgo (School of Physics, Costa Rica Institute of Technology), under the supervision of PhD. Philip Gerlee (Department of Mathematical Sciences, University of Gothenburg & Chalmers University of Technology, Sweden).
This software allows the simulation and classification of Gene Regulatory Networks (GRNs) that drive shape homeostasis in multicellular systems — a key emergent phenomenon in developmental biology and artificial life.
The simulation is entirely written in Python and is based on the framework proposed by Gerlee et al. in the study “The influence of cellular characteristics on the evolution of shape homeostasis”, supported by several foundational models in the field of artificial life cited in the thesis.
This is a fork of the original research project by M.Sc. J. Esteban Pérez-Hidalgo:
Original Repository: https://github.com/j0sees/Simulation-of-Gene-Regulatory-Networks-Driving-Shape-Homeostasis
This fork includes modernization improvements including containerization, web interface, and enhanced documentation while preserving the original scientific research and implementation.
- To simulate multicellular collectives and observe adaptive behavior under dynamic environments.
- To classify GRNs that can maintain shape homeostasis in different environmental conditions.
- To explore the role of growth factors, cell–cell interactions, and internal cellular actions in emergent behavior.
- To demonstrate that three major behavioral classes lead to shape homeostasis.
- To showcase cellular responses as a function of different environmental types.
This software combines a range of well-established computational techniques:
- 🔲 Cellular Automata — for modeling spatial environments and local cell dynamics.
- 🧠 Artificial Neural Networks — to implement gene regulatory networks within each cell.
- 🔍 Optimization Algorithms — to evolve adaptive networks and collective behavior.
These components interact to simulate the dynamics of cellular collectives, where behavior arises from:
- A dynamic equilibrium of cellular actions,
- The interaction between two growth factors enabling communication,
- And a signaling network acting as a gene regulatory network.
- Docker (recommended) - No local Python installation needed
- Python 3.7+ (for local development)
# Clone the repository
git clone https://github.com/j0sees/Simulation-of-Gene-Regulatory-Networks-Driving-Shape-Homeostasis.git
cd simulation-of-gene-regulatory-networks-driving-shape-homeostasis
# Start the web interface (no local Python installation needed)
./run_docker.sh# Clone the repository
git clone https://github.com/j0sees/Simulation-of-Gene-Regulatory-Networks-Driving-Shape-Homeostasis.git
cd simulation-of-gene-regulatory-networks-driving-shape-homeostasis
# Install dependencies
pip install -r requirements.txt# Start the containerized web application
./run_docker.sh
# Open your browser and go to: http://localhost:8501
# Interactive interface with real-time visualizationThe web interface provides:
- Interactive simulation controls with real-time parameter adjustment
- Live visualization of cell states, SGF, and LGF concentrations
- Fitness analysis tools and evolutionary progress tracking
- Parameter exploration for different network architectures
# Run a simple simulation example
python examples/run_simulation.py
# Run genetic algorithm example
python examples/run_evolution.py
# Run main applications directly
python core/main.py [filename] [nNodes] [individual]
python evolution/main_ga.py [filename] [parameters]├── core/ # Core simulation components
│ ├── cell_agent.py # Cell class and neural network
│ ├── main.py # Basic simulation engine
│ └── tools.py # Utility functions
├── evolution/ # Genetic algorithm
│ ├── main_ga.py # Main GA application
│ └── tools_ga.py # GA utilities
├── visualization/ # Plotting and visualization
│ └── plot.py # Plotting functions
├── config/ # Configuration files
│ └── config-ca # NEAT configuration
├── analysis/ # Research tools
│ ├── graph_generator.py # Network generation
│ └── graph_evaluation.py # Network evaluation
├── examples/ # Usage examples
│ ├── run_simulation.py # Basic simulation example
│ └── run_evolution.py # GA example
├── web_app.py # Streamlit web application
├── neat_cell_agent.py # NEAT-based cell agent implementation
├── neat_parallel.py # Parallel NEAT implementation
├── plots_gp # Gnuplot configuration
├── run_web_app.sh # Local web app launcher script
├── Dockerfile # Container configuration
├── docker-compose.yml # Container orchestration
├── run_docker.sh # Container launcher script
├── .dockerignore # Docker build exclusions
├── requirements.txt # Dependencies
├── README.md # Main project documentation
├── docs/ # Documentation
│ ├── web-interface.md # Web interface documentation
│ ├── docker.md # Docker usage guide
│ ├── thesis-summary.md # Research overview
│ └── history-log.md # Development history
core/: Core simulation engine with cellular automata and neural networksevolution/: Genetic algorithm implementation for network optimizationvisualization/: Plotting and visualization toolsanalysis/: Research and evaluation toolsexamples/: Ready-to-run examples for different use casesweb_app.py: Interactive web interface using StreamlitDockerfile: Container configuration for reproducible environments
This software represents a novel, multidisciplinary product, combining:
- Computational Physics - Cellular automata and spatial dynamics
- Numerical Methods in Mathematics - Differential equations and optimization
- Computational Biology - Gene regulatory networks and emergent behavior
It provides a computational platform for studying emergent homeostasis, offering insights into how multicellular systems organize and regulate their shape.
The simulation shows cells in different states (empty, quiet, moving, dividing) alongside SGF and LGF concentration patterns.
Example of a gene regulatory network with feedback loops and regulatory interactions.
Typical fitness progression during genetic algorithm optimization.
- Developmental Biology: Understanding shape formation and maintenance
- Artificial Life: Exploring emergent behavior in cellular systems
- Systems Biology: Modeling gene regulatory networks
- Evolutionary Computation: Optimizing network architectures
- Interactive Web Interface: Real-time simulation control and visualization
- Containerized Deployment: Reproducible environments with Docker
- Modular Architecture: Clean separation of concerns
- Comprehensive Documentation: Detailed explanations and examples
- M.Sc. J. Esteban Pérez-Hidalgo (School of Physics, Costa Rica Institute of Technology)
- Master's Thesis – University of Gothenburg & Chalmers University of Technology
- Email: jose.perez at tec.ac.cr
- Reymer Vargas - Project modernization and development
- Containerization, web interface, and documentation improvements
This project has benefited from contributions by:
- M.Sc. J. Esteban Pérez-Hidalgo - Original research and initial implementation
- Reymer Vargas - Technical Consultant & Platform Engineering Specialist
- Project modernization, containerization, documentation, and web interface development
- Infrastructure as Code, DevOps, and research software engineering
- Master's thesis on gene regulatory networks and shape homeostasis
- Initial implementation of cellular automata and neural networks
- Core simulation engine and genetic algorithm
- Documentation: Comprehensive documentation restructuring and improvements
- Containerization: Docker setup for reproducible environments
- Web Interface: Interactive Streamlit application for real-time simulation
- Code Quality: PEP8 compliance and project structure improvements
- Deployment: Production-ready containerized deployment