lw-pipeline 0.2.0

A lightweight, easy to modify pipeline controller for data processing (e.g. eeg, meg, etc.).

About the Lightweight Pipeline

As the name suggests - a lightweight, easy to modify pipeline. Initially built for EEG analysis using MNE-Python and MNE-BIDS.

Main design criterion is to keep the controller part minimal.

Table of Contents

• Goals
• Usage: rough idea and trivial example
• Minimal example (to use with MNE-BIDS)
• Installation
• Contributing
• License
• Comparison

Goals

• Provide a scheme to model concrete pipeline steps after.
• Take care of a configuration file handling and saving/loading to some extend.
• Decouple the content of a pipeline, i.e. its processing logic, from the organisatorial part.

Usage: rough idea and trivial example

Start by creating a project folder containing:

steps/__init__.py: empty

steps/00_steps.py:

from lw_pipeline import Pipeline_Step

class A_First_Pipeline_Step(Pipeline_Step):
    def __init__(self, config):
        super().__init__("This is a description of a first step.", config)

    def step(self, data):
        print(f"Here data is '{data}'.")
        data = self.config.variable_a
        return data

class A_Second_Pipeline_Step(Pipeline_Step):
    def __init__(self, config):
        super().__init__("This is a description of a second step.", config)

    def step(self, data):
        print(f"Here data is '{data}'.")
        return data

config.py:

steps_dir = "steps/"

variable_a = 1

Now run the following command in the project directory to list the steps detected

lw_pipeline -c config.py --list

while

lw_pipeline -c config.py --run

runs the pipeline entirely. Use --help for further parameter info. You can find a similar example in examples/trivial/.

In a more interesting case, one would pass a data object, e.g. a subclass of Pipeline_Data, through the pipeline. As of now, the pipeline comes with a data container class for processing eeg/meg data utilizing MNE-BIDS.

Minimal example (to use with MNE-BIDS)

For a more interesting example demonstrating eeg/meg processing, we refer to a minimal example.

Installation

To install, clone the github repository, navigate to the folder and execute (for an editable install)

pip install -e ".[dev]"

Contributing

Contributions are welcome! Please fork the repository, create a feature branch, and submit a pull request.

License

This project is licensed under the BSD 3-Clause License. See the LICENSE file for details.

Comparison

• The MNE-BIDS-Pipeline provides an actual eeg/meg processing pipeline that bundles logic and content.