Sample ID C3N-00437 not found in ['./_tutorial_pooled_features/titan/cptac_ccrcc/C3N-00437']

[zs999]

when I run python BAP1_linear_probe.py, I meet this problem. Can you explain under what circumstances this situation might occur? Thanks!

[guillaumejaume]

Hi @zs999,

Do you confirm that patch embeddings AND slide embeddings have been correctly extracted for Sample ID C3N-00437?

[zs999]

I randomly download 6 svs files, which do not include C3N-00437.

Do I need to download all the slide IDs from the CSV file?

[aspartate]

Hi @zs999, thank you for your question. The intended workflow for Pathobench is:

1. Download slides from source URL
2. Use Trident to extract patch features from slides (output: one h5 file per slide_id)
3. Provide the folder of patch features to Pathobench patch_embeddings_dirs
4. Use Pathobench to generate case-level features (one h5 file per case_id) which are saved to pooled_embeddings_dir
5. Use Pathobench to run experiment

In your case, I suspect you provided the wrong directory paths to Pathobench patch_embeddings_dirs and pooled_embeddings_dir. Please ensure that patch_embeddings_dirs is a folder containing patch embeddings and pooled_embeddings_diris an empty folder.

Alternatively, you can extract case-level embeddings using a method of your choice, and save them to pooled_embeddings_dir. Leave patch_embeddings_dirs as None. Then Pathobench will skip step 4 and directly use your pre-extracted case-level embeddings.