(workflows)=

# Workflows

A **workflow** is a pipeline stored in a Git repository. GLACIER currently supports **Nextflow** workflows, detected by the presence of a `nextflow_schema.json` file.

## Installing a workflow

From the Library page, click **Install** on any workflow card. GLACIER clones the repository to your workflows directory at `<configPath>/workflows/<owner>/<repo>@<version>`.

The button label changes from *Install* to *Launch* once the clone completes.

## Configuring parameters

Click an installed workflow card to open the **Parameters** page. This page has three tabs:

### Info
Renders the workflow's `README.md` so you can review documentation before running.

### License
Shows the `LICENSE` file from the workflow repository.

### Params
The main configuration view. GLACIER reads the workflow's `nextflow_schema.json` and renders a JSON Forms form. You can:

- Edit parameter values directly in the form.
- **Load parameters** from a `.json` file on disk.
- **Save parameters** to a `.json` file for reuse.

### Execution profiles

The **Profiles** dropdown lists profiles extracted from `nextflow.config`. These map to Nextflow's `-profile` flag. The `standard` profile is selected by default.

If a `test` profile exists, a **Test launch** button appears alongside **Launch**. Use this to run the workflow with the `test` profile combined with your other selections.

## Resource requirements

Workflow repositories can include a `glacier.json` file at the root to specify minimum CPU and memory requirements:

```json
{
  "resources": {
    "minimum": {
      "cpus": 4,
      "memory": 8
    }
  }
}
```

When you open the Parameters page for a workflow that has a `glacier.json`, GLACIER checks your system against the minimums. A warning banner appears if your effective resources (accounting for GLACIER resource limits, WSL config on Windows, and system hardware) are below the workflow's recommendations. You can still launch the workflow despite the warning.

## Launching a workflow

When you click **Launch**, GLACIER:

1. Validates parameters against the JSON Schema (unless disabled in Settings).
2. Creates a **workflow instance** — a named run directory containing the parameters file, logs, and work directory.
3. Spawns the Nextflow process using the bundled JRE and `nextflow.jar`.
4. Navigates to the **Monitor** page so you can follow progress in real time.

### Launch options

| Option | Behaviour |
|--------|-----------|
| **Launch** | Starts the pipeline with selected profiles and parameters. |
| **Test launch** | Available when a `test` profile exists. Runs with `test` plus any other selected profiles. |

## Resume and restart

- **Resume** (from Monitor page) — re-launches with Nextflow's `-resume` flag, continuing from the last checkpoint.
- **Restart** (from Monitor page) — cancels the running instance and opens the Parameters page so you can edit settings before re-launching.

## What happens after launch

Nextflow is invoked with the following flags:

```
java -jar nextflow.jar run <repo> \
  -work-dir <instance_path>/work \
  -profile <profiles> \
  -params-file <instance_path>/params.json \
  -name <instance_name>
```

Standard output and error are captured to `stdout.log` and `stderr.log` in the instance directory. Nextflow's own log goes to `nextflow.log`.