Missing mappings for connectome bulk files

[psadil]

Several imaging derivatives were contributed to the UKB by Mansour et al., 2023. bioRvix, doi: 10.1101/2023.03.10.532036 and github.com/sina-mansour/UKB-connectomics, and these are available as bulk files/blobs.

https://biobank.ndph.ox.ac.uk/showcase/label.cgi?id=200 https://biobank.ndph.ox.ac.uk/showcase/label.cgi?id=201 https://biobank.ndph.ox.ac.uk/showcase/label.cgi?id=202 https://biobank.ndph.ox.ac.uk/showcase/label.cgi?id=203 https://biobank.ndph.ox.ac.uk/showcase/label.cgi?id=204 https://biobank.ndph.ox.ac.uk/showcase/field.cgi?id=31000

When the relevant fields are specified in ukb-init, they are deposited underneath the sub-[id] directory

$ datalad ukb-init --bids 1009587 20249_2_0 31000_2_0
$ datalad ukb-update --merge -k $UKBKEY
$ ls
31000_2_0  ses-2
$ ls 31000_2_0
MNI_to_T1_inversed_warp_coef.nii.gz

Can these be added to ukb2bids_map.py?

I would be interested enough in this to submit a pull request.

[mih]

A PR would be great!

[psadil]

Are there standards or heuristics for organizing the contents of non-bids? For example, Field 31011 unzips to the following

./31011_2_0:
lh.native.Glasser.annot  native.fMRI_space.Glasser.nii.gz  native.Glasser.nii.gz  rh.native.Glasser.annot

Without additional constraint, I organize this as something like

$ tree ses-2
ses-2
└── non-bids
    ├── fMRI
    │   └── atlas_parcellation
    │       └── native.fMRI_space.Glasser.nii.gz
    └── T1
        └── atlas_parcellation
            ├── lh.native.Glasser.annot
            ├── native.Glasser.nii.gz
            └── rh.native.Glasser.annot

The key choices being

• use of "atlas_parcellation" as a directory
• separating the parcellations by data modality (native.fMRI_space vs native.)
• no attempt to integrate the surfaces into a FreeSurfer directory
• no attempt to enhance the names with BIDS or BIDS-like entities (e.g., ses-2, hemi-lh, space-T1w, etc)

[psadil]

Does that organization seem okay? Let me know if there is a preference for a different amount of integration with the existing structure (that is, either heavier use of BIDS entities or something closer to an unzipping of subfolders by field within the non-bids subdirectory).