Lukas Heumos
Lukas Heumos
Uhm, interestingly enough there is a version on pypi that is higher than the version in the setup.py here. Not sure what's happening. The latest release on pypi was still...
Ahh, thank you! Would you be interested in playing around https://github.com/scverse/pertpy-tutorials/blob/main/milo.ipynb and see whether you can make it use the non-iterative method?
Dear @FrenkT have you been able to find something?
No worries! > Increasing neighborhood size moves things in the right direction, but even with n_neighbors=300 the milo count matrix is still quite sparse. that's also a huge amount of...
It does but only if the solver is `edger` (see just a few lines above). This is expected behavior, right? We are mentioning rpy2 here: https://github.com/scverse/pertpy?tab=readme-ov-file#milo but I see that...
Ah, I missed that, sorry. I'm only partially a fan of an extra named `milo` because it could imply that milo doesn't work if you don't have that extra installed....
> would you be open to moving pertpy from the conda-forge to the bioconda channel for an optional dependency? No. We actually moved all scverse packages from bioconda to conda-forge...
@brainfo thank you for your suggestion! I'll have a proper look soon.
> Does Theislab not have any servers? I can run it on some server I have access to otherwise We have plenty of compute of course but it's difficult to...
Are they planning to format yaml?