🧬 GenoFlow: The Visual IDE for Genetic Regulatory AI

** Note this application is still in development. Please pin or note bugs.

GenoFlow (formerly Genome Studio) is a visual, node-based development environment for designing, training, and tuning GENREG (Genetic Regulatory) AI models.

Inspired by biological gene regulatory networks, GenoFlow lets users visually construct complex, self-adapting AI architectures that combine a functional neural network layer with a stateful, trust-modulating protein network.

---

Key Features

Visual, Node-Based Architecture

Build and visualize your entire AI model using an intuitive, ComfyUI-like graph editor powered by LiteGraph.js.

Biologically Inspired Layers (GENREG)

Protein Network (Regulatory Layer) Stateless, self-adapting proteins (e.g., Sensor, Trend, Comparator, TrustModifier) process environmental signals to generate a Trust Delta (fitness signal).

Controller Network (Functional Layer) A simple feed-forward neural network that selects actions based on processed signals.

Trust-Based Evolution

Genome fitness is determined by accumulated Trust, promoting robust and adaptive behaviors through evolutionary pressure.

Real-Time Environment

Train and observe AI behavior in a simulated Snake Environment, with real-time visualization via a separate Pygame window.

Client–Server Architecture

A Python FastAPI / WebSocket backend handles all heavy processing (evolution, environment steps), while the JavaScript frontend provides the interactive IDE.

---

Quick Start

1. Prerequisites

GenoFlow requires Python and several libraries, including fastapi and uvicorn for the server, and optionally pygame for environment visualization.

# Recommended: Create and activate a virtual environment
python -m venv venv
source venv/bin/activate

2. Installation

Install the required Python packages:

pip install -r requirements.txt

# To enable environment visualization
pip install pygame

Note requirements.txt includes core dependencies such as fastapi, uvicorn, and websockets.

3. Running the Server

Start the backend server using uvicorn as defined in start_server.py:

python start_server.py

The server typically starts at:

http://0.0.0.0:8000

4. Accessing the IDE

Open your browser and navigate to:

http://localhost:8000

Load the provided snake_training_template.json to begin exploring the training flow.

---

Architecture Overview

GenoFlow is divided into distinct layers reflecting its biological inspiration and client–server design.

Component	Technology	Role	Core Files
Frontend IDE	LiteGraph.js, Vanilla JS	Visual graph editor, monitoring, real-time control	static/js/*.js, static/index.html
Backend Server	Python (FastAPI, WebSocket)	AI processing, evolution, simulation	start_server.py
Controller Layer	Python (Controller class)	Feed-forward NN for action selection	genreg_controller.py
Regulatory Layer	Python (Protein classes)	Generates Trust Delta fitness signals	genreg_proteins.py
Evolution Core	Python (Genome, Population)	Evolutionary process and selection	genreg_genome.py, genreg_population.py

---

The GENREG Model Flow

The training process is orchestrated by the node graph in a continuous loop:

1. Environment Step (SnakeEnvironment) outputs signals.
2. Protein Network processes signals and calculates Trust Delta.
3. Controller Network selects an Action.
4. Action is fed back into the environment.
5. Episode ends and total Trust determines Genome fitness.
6. Generation Manager triggers population evolution.

---

Node Types

The IDE provides modular nodes to construct and monitor AI behavior.

Category	Example Nodes	Purpose
Regulatory	Sensor, Trend, Trust Modifier	Process signals and influence Genome Trust
Functional	Controller Network	Select actions from processed signals
Environment	Snake Environment, Visualize (Pygame)	Simulate and visualize behavior
Evolution	Population Controller, Episode Runner	Manage training and mutation lifecycle

---

Contribution

GenoFlow is an evolving project. Contributions are welcome.

Areas of interest:

• New protein types
• Additional environments
• Visualization tools
• Extensions to evolutionary algorithms

Pull requests and design discussions are encouraged.