The 'sklearn' PyPI package is deprecated, use 'scikit-learn' error

debbiemarkslab / EVcouplings

Evolutionary couplings from protein and RNA sequence alignments

http://evcouplings.org

Other

232 stars 75 forks source link

The 'sklearn' PyPI package is deprecated, use 'scikit-learn' error #296

Closed fbalo-medium closed 2 months ago

fbalo-medium commented 1 year ago

I sometimes get this error when installing evcouplings via pip.

ChristophKnapp commented 1 year ago

I have the same problem. Allthough its not sometimes for me.

Seems to me we need to dumb down the python version until sklearn is not deprecated anymore or sklearn imports need to be replaced with scikit-learn. I would prefer the later.

Is it better to install the development version?

ChristophKnapp commented 1 year ago

The development version installs without problems for me.

thomashopf commented 1 year ago

Yes, please use the development branch for now while we are in the process of preparing a new public release. You may need to use an older version of numpy and pandas if there are any error messages.

zachcp commented 1 year ago

+1 for updating to newer numpy and updated release. Are there tasks you could use help on?

fbalo-medium commented 12 months ago

For who might need this; I was able to downgrade the packages below to this versions for the development branch to work for me.

pip install pandas==1.2.4 seaborn==0.11.2 scipy==1.7.1 numba==0.53.1 matplotlib==3.4.1 numpy==1.19.5

joelmeyerson commented 3 months ago

@thomashopf Is there still a new public release in the works?

thomashopf commented 3 months ago

Yes - the v0.2 branch is a good step closer to the new release ... I should be able to finalize this in the next weeks hopefully

thomashopf commented 2 months ago

The latest develop branch version should fix the sklearn/scikit-learn issue - and if it behaves stably during testing for a few weeks, it will become v0.2 eventually.

Note this also should fix the numpy/pandas versions by replacing all deprecated function calls. If you catch any others running the most recent version, please reopen the issue.