There is a bug in the Scrublet wrapper I contributed to Scanpy, which does seem quite important and was creating some weird results in my testing.
I've submitted a fix for that, and at the same time proposed the relocation of batch-handing code from here to there. I was going to need to re-write the Scrublet plotting function anyway, to deal with batches since I was having trouble wrapping the outputs here in a batch-aware way, so it made sense just to improve the upstream Scanpy code while I was fixing the bug.
I've therefore cut the batch handling out of this module again, and hopefully it can be handled by the upstream code once the Scanpy people approve. I will add a patch to our Bioconda recipe to apply a patch to add that functionality, pending a release (which can be a while with Scanpy).
pinin4fjords
commented
2 years ago
There is a bug in the Scrublet wrapper I contributed to Scanpy, which does seem quite important and was creating some weird results in my testing.
I've submitted a fix for that, and at the same time proposed the relocation of batch-handing code from here to there. I was going to need to re-write the Scrublet plotting function anyway, to deal with batches since I was having trouble wrapping the outputs here in a batch-aware way, so it made sense just to improve the upstream Scanpy code while I was fixing the bug.
I've therefore cut the batch handling out of this module again, and hopefully it can be handled by the upstream code once the Scanpy people approve. I will add a patch to our Bioconda recipe to apply a patch to add that functionality, pending a release (which can be a while with Scanpy).