conda version is out of date.

[bioworker]

Celltypist in conda is out of date (2023-2-9), methods like extract_top_markers is not in the conda version. Thanks for the nice work.

[ChuanXu1]

@bioworker, as I have tested, conda will install the newest version. You can paste your installation information here if you still fail to install the newest CellTypist.

[bioworker]

@bioworker, as I have tested, conda will install the newest version. You can paste your installation information here if you still fail to install the newest CellTypist.

Ok, The version information suggested the newest one (1.3.0), but it lack the methods like extract_top_markers, did you check this method? Now I have reinstalled it by pip command, everything is fine now.

[prete]

Can confirm that latest conda-installed version is working.

Instaled using:

conda install -c bioconda -c conda-forge celltypist

Then:

>>> import celltypist
>>> celltypist.__version__
'1.3.0'
>>> model = celltypist.models.Model.load(model = 'Immune_All_Low.pkl')
>>> model.extract_top_markers("Mast cells", 3)
array(['CPA3', 'TPSAB1', 'FTH1'], dtype=object)

[benemead]

Hi there - I also encountered this issue, but I believe it's Mac apple silicon dependent. when I run: conda install -c bioconda -c conda-forge celltypist

My install completes, but to 0.1.9.

However, when I run: conda install -c bioconda -c conda-forge celltypist=1.3

I get the following solve error:

Output in format: Requested package -> Available versionsThe following specifications were found to be incompatible with your system:

  - feature:/osx-arm64::__unix==0=0
  - celltypist=1.3 -> click[version='>=7.1.2'] -> __unix
  - celltypist=1.3 -> click[version='>=7.1.2'] -> __win

Your installed version is: 0

FYI - pip install works fine

[prete]

Indeed it seems to be a Mac M1/M2 issue. The arm architecture probably has conflicting packages.

To solve this issue they can try rosetta. Useful links here and here. Rosetta should translate intel instructions and if they are using conda, they should make sure that the right conda installer (ARM version) is used. The info in this issue can also help.

Otherwise, and as you mentioned, sticking with pip should not give you any issues.