Enable parsing of BIDS dataset with data dictionary

[surchs]

This tracks the work to create a bagel add-bids subcommand that does this:

Inputs:
(1) pheno.jsonld generated by bagel pheno
(2) BIDS dataset directory path
(3) maybe: the path on the host computer of the BIDS dataset
Output: pheno_bids.jsonld

A user who has already run bagel pheno (#32) successfully to create a pheno.jsonld file should be able to use this subcommand to add imaging metadata (BIDS) to the existing pheno.jsonld file. The imaging metadata to add are:

• the type of imaging acquired during a session as Acquisition
• the absolute file system path of a session directory in the BIDS dataset

Expectations (for now):

• subject IDs in the pheno.tsv and the corresponding BIDS directory will be the same.
• there may be subjects in the pheno.tsv file that do not exist in the BIDS directory (i.e. pheno only)
• if a subject does not exist in the BIDS dir, it will remain in the .jsonld with only the existing phenotypic metadata
• if a subject has multiple acquisitions of the same type in the same session (e.g. several T1s), then they should be added as separate acquisitions to the session (i.e. you can have more than one T1 in a session)

Things to do:

• add a new subcommand for parseing BIDS data #76
  • this will require refactoring the existing bids pheno CLI tests because the CLI will have to be called with the explicit subcommand name whereas we can currently just call it with the cli name (i.e. bagelbids --somearg vs bagelbids pheno --somearg)
• ensure we can create a datamodel instance from a pheno.jsonld input #77
  • the pheno.jsonld is currently only valid after we remove the @context tag
  • we can investigate if there is a way to adjust the data dictionary schema to also allow for the @context while still rejecting invalid data dictionary instances. but stripping the@context is also a valid solution for now
• create representation of BIDS dataset #29
• add BIDS metadata to existing phenotypic metadata #79
  • essentially matching subjects by label and adding new imaging related attributes without deleting the existing information (or changing the subject identifier / reinstantiating any of the dataset / subject / sessions instances)
• For BIDS subjects with only one session (and therefore no ses subdirectory), create a new session object with label 01 #82
• Encode the absolute file path #4
  • get the absolute file system path, even if the path to the BIDS directory was provided as a relative path
  • add file path attribute to model schema for a session
• Output the combined pheno + BIDS .jsonld #83
  • add the @context back in, if it was previously stripped
  • make sure the existing phenotypic metadata is still there / the same / unchanged
  • make sure the BIDS imaging data matches our expectations from looking at the input data

[alyssadai]

@surchs I think this one can be closed as well, but want a second pair of eyes before doing so.

[surchs]

@alyssadai agreed! Let's close 🎉