conversion to SBU format is broken

[kaczmarj]

At the moment, this error is raised, and I assume there will be other errors as well if that one is fixed:

  File "/xxxxxxx/wsinfer/cli/convert_csv_to_sbubmi.py", line 342, in tosbu
    class_names: list[str] = run_metadata["model_weights"]["class_names"]
KeyError: 'model_weights'

[kaczmarj]

@swaradgat19 - could you please look into this?

[kaczmarj]

the conversion script also needs to be updated to glob multiple image file types, not just SVS files.

[swaradgat19]

Sure. I'll look into it

[swaradgat19]

@kaczmarj Could you tell me a little about execution-id, study-id and make-color-text fields in the tosbu command?

[kaczmarj]

kaczmarj Could you tell me a little about execution-id, study-id and make-color-text fields in the tosbu command?

of course, i will be honest, i am a bit fuzzy about this format. i learned about the format by reviewing the structure of several existing files.

• execution-id - a unique ID for the run. I believe this is a UUID (which you can create in python using the uuid module in the standard library).
• study-id - this is a string that identifies the group of people or samples that were analyzed. this is an arbitrary string, like TCGA-BRCA).
• make-color-text - the format includes two files. one of the files is a text file that contains color information of each patch. i am not sure if this color file was always required, so i left it as an option. the creation of this color text file can take a considerable amount of time because the patches need to be loaded from the whole slide image.

let me know if you have any other questions.

[swaradgat19]

Got it. So if I want to run this tosbu command, what should I put as the execution-id?

[kaczmarj]

for now, use a UUID. i show how to make one in python below:

import uuid
execution_id = str(uuid.uuid4())

[kaczmarj]

closed in favor of #113