scanpy: scikit-misc cannot be initiated when sc.pp.highly_variable_genes(adata, n_top_genes=5000, flavor='seurat_v3')

I have checked that this issue has not already been reported.
I have confirmed this bug exists on the latest version of scanpy.
(optional) I have confirmed this bug exists on the master branch of scanpy.

Hello Scanpy, When I’m running sc.pp.highly_variable_genes(adata, n_top_genes=5000, flavor=‘seurat_v3’), it asks me to install scikit-misc, which is already installed. Please see the picture below. Could you please help me to solve this issue? Thanks! Best, YJ Note: Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

Minimal code sample (that we can copy&paste without having any data)

# Your code here

[Paste the error output produced by the above code here]

Versions

3.8

[Paste the output of scanpy.logging.print_versions() leaving a blank line after the details tag]

scanpy==1.8.2 anndata==0.7.8 umap==0.5.2 numpy==1.20.3 scipy==1.7.2 pandas==1.3.4 scikit-learn==1.0.1 statsmodels==0.13.1 python-igraph==0.9.8 pynndescent==0.5.5 scvelo==0.2.4 scanpy==1.8.2 anndata==0.7.8 loompy==3.0.6 numpy==1.20.3 scipy==1.7.2 matplotlib==3.5.0 sklearn==1.0.1 pandas==1.3.4 cellrank==1.5.0 scanpy==1.8.2 anndata==0.7.8 numpy==1.20.3 numba==0.54.1 scipy==1.7.2 pandas==1.3.4 pygpcca==1.0.2 scikit-learn==1.0.1 statsmodels==0.13.1 python-igraph==0.9.8 scvelo==0.2.4 pygam==0.8.0 matplotlib==3.5.0 seaborn==0.11.2

About this issue

Original URL
State: closed
Created 3 years ago
Comments: 19 (3 by maintainers)

Most upvoted comments

Had this problem, followed the scikit-misc package issue on a related problem and installed the recommended patch with

pip install -i https://test.pypi.org/simple/ "scikit-misc==0.2.0rc1"

Seems to work now for me.

allumik on Mar 29, 2023

I found a workaround that does not require downloading the .whl file for numpy=1.19.5. By default, MKL is included when you install numpy with conda. It’s good to do this in a new environment.

conda create -n scanpy_env
conda activate scanpy_env
conda install numpy=1.19
conda install seaborn scikit-learn statsmodels numba pytables
conda install -c conda-forge python-igraph leidenalg
pip install scanpy==1.8.1

Now I can run sc.pp.highly_variable_genes() with no problem.

ghost on Mar 3, 2022

Hello @davidhbrann, Thanks for the suggestion. I found that numpy‑1.19.5+mkl‑cp36‑cp36m‑win_amd64.whl only supports python 3.6, but the support of python 3.6 has been dropped by Scanpy. I didn’t find numpy-1.20.3+mkl-cp38-cp38-win_amd64.whl by google yet, but I believe that your solution will work. It seems that numpy‑1.21.5+mkl‑cp38‑cp38‑win_amd64.whl is the only hope, but Scanpy requires numpy<=1.20. I think it’s the compatibility issue. Hope Scanpy can solve this compatibility issue in the future. Thanks! Best, YJ

hyjforesight on Dec 29, 2021

I met the same problem one day ago. I run the following: conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch conda install scvi-tools -c conda-forge conda install -c conda-forge scanpy python-igraph leidenalg It works fine now.

aqlkzf on Sep 27, 2022

I found a workaround that does not require downloading the .whl file for numpy=1.19.5. By default, MKL is included when you install numpy with conda. It’s good to do this in a new environment.
conda create -n scanpy_env
conda activate scanpy_env
conda install numpy=1.19
conda install seaborn scikit-learn statsmodels numba pytables
conda install -c conda-forge python-igraph leidenalg
pip install scanpy
Now I can run sc.pp.highly_variable_genes() with no problem.

Update: this workaround does not seem to work anymore, at least for scanpy 1.8.2 (you’ll need to pip install scanpy==1.8.1). During pip install scanpy, a newer version of numpy is installed and version 1.19 is overwritten. This newer version does not have MKL, leading us back to square one. It’s also not possible to conda install numpy 1.19 as the very last step, because this leads to another error (it’s related to the fact that scanpy needs to be compiled with the same version of numpy).

ghost on Mar 3, 2022