Change stained microscopy directory schema to be more generic and not specific to PAS microscopy, use OME-TIFF as raw data format

[NHPatterson]

• Does this change have buy-in from other groups providing the same kind of data?

This should expand the file types of raw data that can be uploaded. Right now, the specification only allows .scn files to be uploaded which are from Vanderbilt TMC's early specification. We at Vanderbilt TMC now do this scanning with a Zeiss instrument that creates .czi files. Although for older data we may still have .scn files so I don't recommend a total deprecation of this extension.

• Can you quote or reference any outside documentation that motivates this change?

We'd like to upload our data to HuBMAP.

• Is this change backward compatible, or does it make validation more stringent?

This should be backwards compatible if the acceptable file extensions are expanded.

@icaoberg will evaluate this request, and if it can move forward, will file a PR for @pdblood to review, and @mccalluc will make sure tests pass before merging.

[NHPatterson]

Changing direction to be more general

After speaking with @pdblood, we realized this issue may have broader implications.

• The directory schema is listed as "PAS Stained Microscopy" but the metadata.tsv has a stain field where one can select PAS or H&E or another histological stain. Going forward, this should be generalized to "Stained Microscopy" and be an umbrella for H&E, PAS, and other histological stains that are usually imaged by brightfield microscopy.

• The raw data should be less vendor focused. In version 1, the schema checks for .scn and we initially proposed to add .czi to accommodate us, Vanderbilt, specifically as we capture these images on a new instrument. A better idea would be to upload any raw data as a losslessly converted OME-TIFF as well, making the level 0 data more FAIR. This is aided by the fact that there are now several easy to use OME-TIFF conversion tools.

OME-TIFF conversion tools:

Command-line

bioformats2raw -> raw2ometiff

Probably the tool of reference for this task, uses command line to perform conversion. 2-step process. bioformats2raw raw2ometiff

GUI

QuPath

GUI interface, users can open images in QuPath and convert through file menu options. QuPath

Kheops

GUI interface through FIJI imageJ. 2-step install process through FIJI update sites. Kheops

[NHPatterson]

Is it possible to resolve this? The current implementation's raw data is vendor locked to Leica .scn files that we no longer create in our lab.