XLSX derivative files.

[TheChymera]

BIDS does not support *.xlsx files.

There are a number of such files in the derivatives:

derivatives/MBF_stereology/sub-I48/sub-I48_sample-BrocaAreaS30_stain-NeuN_loc-Layer6_STER.xlsx
derivatives/MBF_stereology/sub-I48/sub-I48_sample-BrocaAreaS30_stain-NeuN_loc-Layer3_STER.xlsx
...
derivatives/MBF_stereology/sub-I46/sub-I46_sample-BrocaAreaS11_stain-Calretinin_loc-Layer3_STER.xlsx
derivatives/MBF_stereology/sub-I46/sub-I46_sample-BrocaAreaS22_stain-NeuN_loc-Layer3and5_STER.xlsx

Ideally they should be dealt with in one of the following manners:

• reformatted to .tsv or .json
• deleted

[satra]

i'm not sure it doesn't support it under derivatives. but indeed should be checked if these could be converted to tsv.

[TheChymera]

The string xlsx is nowhere to be found in the BIDS schema, nor in the whole specification. Perhaps it's implied somewhere, and please correct me if it is.

[satra]

derivatives don't have the same restrictions as "rawdata" for either names or types as far as i know. thus technically you could store anything in any name as long as the pipeline and dataset type for the derived dataset was covered in a dataset_description.json.

see:

• https://bids-specification.readthedocs.io/en/stable/03-modality-agnostic-files.html#derived-dataset-and-pipeline-description
• https://bids-specification.readthedocs.io/en/stable/02-common-principles.html#storage-of-derived-datasets
• https://bids-specification.readthedocs.io/en/stable/05-derivatives/01-introduction.html#bids-derivatives

for various derivatives standardization is going to take a while, so best practice is going to be more important than invalid. specifically see this section:

• https://bids-specification.readthedocs.io/en/stable/02-common-principles.html#non-compliant-derivatives

[TheChymera]

ok, I stand corrected. In that case we should either be extra demanding and require BIDS-style derivatives only, or just forego validation for derivatives. This also means that most of my new issues can technically be disregarded, and I should just make sure that the derivatives directory is skipped.

[satra]

i think we should still check the derivative dataset convention (e.g., dataset_description existence and fields). the naming convention is a bit moot at this point but perhaps we could build up warnings for them.