7.11. Creating a Data Registry

7.11.1. What is a data registry

Data Registry is an object which manages pipeline data (like parcellations, coordinates and masks) based on the scope. junifer comes with the following in-built data registries:

• ParcellationRegistry for parcellations

• CoordinatesRegistry for coordinates

• MaskRegistry for masks

You interact with them indirectly via get_data(), load_data(), list_data(), register_data() and deregister_data(). The kind parameter in them directs which data registry to interact with. These in turn supply preprocessors and markers with their necessary data for computation.

7.11.2. How to make a data registry

Ideally you would not need to create a custom data registry if you work with fMRI data. In case you work with other modalities like EEG, you might want to create a data registry for montages. Here is how you would go about it:

1. Check extending junifer on how to create a junifer extension if you have not done so.

2. Create the data registry in the extension script like so:

   from junifer.api.decorators import register_data_registry
   from junifer.data import BasePipelineDataRegistry
   
   
   @register_data_registry("montage")
   class MontageDataRegistry(BasePipelineDataRegistry):
       def __init__(self):
           super().__init__()
   
       def register(self):
           pass
   
       def deregister(self):
           pass
   
       def load(self):
           pass
   
       def get(self):
           pass
   

   • register_data_registry() registers a class with the name passed in the argument, "montage" in this case.

   • Inheriting from BasePipelineDataRegistry takes care of the class acting as a data registry.

   • BasePipelineDataRegistry.register(), BasePipelineDataRegistry.deregister(), BasePipelineDataRegistry.load() and BasePipelineDataRegistry.get() need to be implemented (check other registries for reference implementations).

3. Pass kind="montage" in *_data functions to verify if your data registry is set up properly.