Experiment Setup

Before running any analysis, you need to create an experiment. This sets up the directory structure and configuration file that the pipeline uses.

What you provide

Field	Description
Work Directory	Parent folder where experiment folders are created
Experiment Name	Name for your experiment (becomes a subfolder)
Data Directory	Folder containing your input segmentation files
Config Template	A YAML configuration template
Cores	Number of CPU cores for parallel processing

The New Experiment button enables once all fields are filled in and the config template is a .yml or .yaml file.

Creating a new experiment

1. Choose a Work Directory.
2. Enter a unique Experiment Name.
3. Select your Data Directory.
4. Pick a Config Template File. The template is read and the UI (including segmentation values) is populated so you can review and edit before creation.
5. Set the number of Cores.
6. Review the Segmentation Values — these are populated from the template but should be edited to match your data.
7. Click New Experiment.

What happens under the hood

• A folder named after your experiment is created inside the work directory.
• The config template is copied into that folder as <experiment_name>_config.yml.
• The copied config is populated with your settings: data_dir, work_dir, exp_name, cores, segmentation_values, and script_location.

Config template location matters

The GUI automatically detects the location of the analysis scripts based on where your config template file is located. Keep your config template inside the surface_morphometrics directory (the repository containing the pipeline scripts). If you move the config file elsewhere, the GUI will not be able to find the scripts.

Segmentation values

The segmentation values section lets you define label–value pairs for your membrane classes. These are written into the config file and used throughout the pipeline.

When you select a config template, the GUI auto-populates these values from the template's defaults. Review and update them to match your specific dataset before creating the experiment.