First off, thank you for this excellent benchmark — it's been incredibly valuable for the single-cell community to have a systematic comparison of SSL methods with such thorough evaluation. Really appreciate the work you've put into this!
I'm trying to reproduce and build on the results, but I've hit a roadblock with the data. The prepare_dataset.py functions reference preprocessed files (under /home/baunsgaard/scBench/scButterfly/Olga_Data/) that don't seem to be publicly available. I was able to find most of the raw public sources, but the exact preprocessed versions used in the benchmark would save a lot of time and guesswork.
Would it be possible to upload the preprocessed dataset files somewhere (Zenodo, Figshare, etc.)? I know uploading data is never as straightforward as it sounds, but it would mean a lot to anyone trying to build on your benchmark.
Thanks again for all your hard work — it's a great resource for the field!
First off, thank you for this excellent benchmark — it's been incredibly valuable for the single-cell community to have a systematic comparison of SSL methods with such thorough evaluation. Really appreciate the work you've put into this!
I'm trying to reproduce and build on the results, but I've hit a roadblock with the data. The prepare_dataset.py functions reference preprocessed files (under /home/baunsgaard/scBench/scButterfly/Olga_Data/) that don't seem to be publicly available. I was able to find most of the raw public sources, but the exact preprocessed versions used in the benchmark would save a lot of time and guesswork.
Would it be possible to upload the preprocessed dataset files somewhere (Zenodo, Figshare, etc.)? I know uploading data is never as straightforward as it sounds, but it would mean a lot to anyone trying to build on your benchmark.
Thanks again for all your hard work — it's a great resource for the field!