# Web GUI Quickstart (Standalone)

The bundled web GUI is a visual DAG editor for building pyCyto pipelines without writing YAML by hand — designed for biologists and other non-programmer users.

This guide covers **standalone mode**: running everything locally via Docker Compose, with no cluster/SLURM connection. If you need cluster execution, see your institution's advanced setup guide.

## 1. Start the stack

```bash
docker compose -f docker-compose.distribution.yml up
```

Open `http://localhost:3003` in your browser.

## 2. Build a pipeline

1. Drag nodes from the palette (left) onto the canvas — organized by stage: Data I/O, Preprocessing, Segmentation, Tabulation, Tracking, Postprocessing.
2. Connect nodes to define the DAG (data flows top-to-bottom by pipeline stage).
3. Click a node to open the **Inspector** (right panel) and set its parameters (e.g. channel names, model thresholds).
4. Use the **File Browser** to point Data I/O nodes at your input images, mounted into the container's `/data` volume.

## 3. Compile and run

1. Click **Export** to compile the DAG into a `pipeline.yaml` / `pipeline-resources.yaml` pair, written under `/data/output/webgui_pipelines`.
2. Run the compiled pipeline from a terminal against the `cyto` service:
   ```bash
   docker compose -f docker-compose.distribution.yml run --rm cyto \
     cyto --pipeline /data/output/webgui_pipelines/<workflow>.yaml -v
   ```

The GUI's in-app **Run** button currently submits to a SLURM cluster and isn't available in standalone mode — use the Export + CLI steps above until local execution is added to the web GUI.

## 4. Save / load

Use **Save** / **Load** in the top bar to persist a DAG design as a workflow file you can reopen later or share with a colleague.

## Notes

- Standalone mode has no SLURM/cluster connection — pipeline execution happens via the `cyto` CLI as shown above.
- For large datasets, GPU-accelerated segmentation (StarDist/Cellpose) requires the NVIDIA Container Toolkit — see [docker-install](docker-install.md).