pytc-client

Recommended Setup (uv)

We strongly recommend using uv because it bundles Python 3.11, installs dependencies in seconds, and keeps the workflow to two commands.

Requirements (uv)

• git
• uv
• Node.js 18+ (includes npm)

1. Bootstrap (one-time)

./scripts/bootstrap.sh                  # macOS / Linux
scripts\bootstrap.ps1                   # Windows PowerShell

2. Launch the full stack

./start.sh                              # macOS / Linux
start.bat                               # Windows CMD

This starts both FastAPI services under uv, runs the Electron client, and cleans up the servers when you close the UI. Re-run the bootstrap script any time you need to refresh dependencies.

Containerized Backend (Docker)

Prefer to keep everything in a container? The provided image uses uv inside the container so the backend is ready in one build step. This target runs the FastAPI API (:4242), the PyTC worker service (:4243), the Neuroglancer bridge (:4244), and TensorBoard (:6006). Electron is not included—use your local client or Electron build to talk to the containerised backend.

Requirements (Docker)

• Docker with Compose V2 (docker compose)

Build the image

docker compose build backend

Run the backend services

docker compose up backend

When you are done:

docker compose down

You can now point the desktop client at http://localhost:4242.

Legacy Manual Setup (without uv)

If you prefer to manage Python environments yourself, follow the steps below. Expect a slower install and more moving pieces.

Requirements (manual)

• Python 3.9–3.11 (conda, pyenv, or system Python)
• git
• Node.js 18+ (includes npm)

1. Create / activate an environment

   python -m venv .venv
   source .venv/bin/activate        # Windows: .\.venv\Scripts\activate

2. Install backend dependencies

   pip install -r server_api/requirements.txt

3. Download & install pytorch_connectomics

   ./setup_pytorch_connectomics.sh
   pip install -e pytorch_connectomics

4. Install frontend dependencies

   cd client
   npm install
   cd ..

5. Run the app (three terminals or background processes)

   python server_api/main.py
   python server_pytc/main.py
   cd client && npm run electron

Video Demo

Video walkthrough