Efforts migrated to: bids-standard/pybids#724
For now this is a prototype to explore the possibility of creating self-contained tool to perform the data aggregation necessary to jump from BIDS events.tsv files/model.json/other time series metadata to a design matrix for downstream implementers (or an appropriate sparse precursor of the design matrix).
Design notes: https://hackmd.io/QdwXR8XwRcmZaXp1zw6Ukg
- Free software: MIT license
- Documentation: https://bids-statsmodels-design-synthesizer.readthedocs.io.
Developer setup --------While this tool attempts to have no dependencies, the development dependencies for now can be installed with (very hacky for now): :: # you may need to brew install git-annex # first step is to install git-annex with something like: brew install git-annex conda create -c conda-forge -n bids-stats-synth python=3 conda activate bids-stats-synth
pip install datalad pip install -r requirements_dev.txt pip install -e .
mkdir tests/data cd tests/data datalad install ///openneuro/ds000003 mkdir ds000003/models curl -fsSL https://raw.githubusercontent.com/poldracklab/fitlins/master/examples/models/ds000003/models/model-001_smdl.json > ds000003/models/model-001_smdl.json
cd .. pytest
Works for at least one test case!
- Try to get some transformations working without dependencies.
- Deal with the absence of dependicies gracefully.
- Get a single file implementation built with https://github.com/Akrog/pinliner.
- Drop boutiques.
- Write up a description of the transformation spec.
- Translation to formats easy for other implmenters to injest.
- Maybe a way to stop prior to densifying transformation.
This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.