The tool I use now is slow, unvalidated and the results look weird sometimes.
I previously discarded scrublet, which has a preprint on biorxiv, but after discussion with the author, I could resolve the issues.
The problem was that simulated doublets works only well when running the tool on a single sample and not the entire dataset, as doublets cannot be formed from different samples.
The tool I use now is slow, unvalidated and the results look weird sometimes.
I previously discarded
scrublet, which has a preprint on biorxiv, but after discussion with the author, I could resolve the issues.The problem was that simulated doublets works only well when running the tool on a single sample and not the entire dataset, as doublets cannot be formed from different samples.
--> switch back to scrublet.