Quentin Blampey

> Yes, but according to the log-file they only filtered patches due to transcript counts. But it seems as this is due to the data itself. This is also fine,...

Could you maybe try without the Dask parallelization backend, i.e., running the patches in sequential mode? I know it will be slow, but I want to see if the issue...

Thanks for trying @jonas2612! Btw, you can also try [nf-core/sopa](https://nf-co.re/sopa/dev/) if you are familiar with Nextflow. If the issue is related to the dask backend, then the nextflow pipeline should...

Hi @brianysoong, Just to clarify: do you have any error running stardist? Or do you only have warnings, but it runs successfully? Concerning the artefacts, this is due to hallucination,...

Hi @hallumg! Did `sopa explorer write` re-created a new image or just made a symlink to the original image? 1. If it created a new image, can you check the...

Hi @hallumg, sorry for the delay This error is surprising; it seems unrelated to the mode because it fails when opening the `tuto.zarr` SpatialData object. Are you sure this `tuto.zarr`...

Thanks @hallumg, please let me know! I know it can work with the version 4 of the Explorer, but it may be indeed less stable, I'll also investigate more on...

Hi @Mr-Milk, Thanks for sharing this, it looks really interesting! If I understand correctly, this could be directly added as an aggregation function, rather than creating an intermediate image for...

Hello @ccruizm, There is not a 1-to-1 mapping between the prior cells and the proseg cells, because one prior cell may be split into multiple proseg cells, and, conversely, multiple...

Thanks for the information @ccruizm, I didn't know that. In that case, I'm a little bit surprised we don't have the info, because we read all the metadata from Proseg's...