David S. Fischer

Hi @niklaslang thanks for the issue, could you post a full or slightly longer error trace, please?

Thanks @Zethson I would guess it s about dask vs sparse version then looking at the error trace.

can you install and import other packages @helloitsmeteeheehee ? i would suspect this is an evironment issue that could be diagnosed in this way.

> should i use the general root directory for pip install -e? or should i be more specific and put it into a subroot (ie: Users)? The directory does not...

Sorry for the delay, I will check whether this still occurs in the new version.

Hi @michalk8, this is on the horizon now as we phase in new linear algebra backends, I will keep this issue open as we go!

Hi @grst, I assume `n_genes` is numeric? We enforce that numeric covariates are independently verified to be desired numeric and not by accident converted from one-hot categorical. This is done...

Hi @dburkhardt! Thanks for the issue! I ll try to rephrase your scenario: Do you want to test each cluster against all other clusters? Or each cluster against every other...

I just improved the zero variance handling on the dev branch. I noticed that the problem that you encountered should have been caught by an update that I pushed to...

The necessary versions of batchglm and diffxpy are now on dev and master each and on pypi.