Input data structure

[tsalo]

Unlike fMRIPost-AROMA or fMRIPost-tedana, this pipeline is going to require both the raw BIDS dataset (for the phase data) and the fMRIPrep derivatives dataset (for the transforms and preprocessed magnitude data).

One thing that I think might be tough will be linking preprocessed magnitude data to the original phase data. Hopefully it's just a matter of BIDSLayout.get_nearest using the preprocessed magnitude file, while explicitly dropping any derivatives-only entities, but I'm not optimistic.

[tsalo]

We might just want to go preprocessed magnitude --> raw magnitude --> raw phase.