search

spacetx / starfish

starfish: unified pipelines for image-based transcriptomics
https://spacetx-starfish.readthedocs.io/en/latest/
MIT License
225 stars 67 forks source link

Couldn't install without specifying an older version of scikit-image (0.17.2) #1988

Closed ryandikdan closed 2 years ago

ryandikdan commented 2 years ago

Hello, I have been trying to install starfish and have come across an error. I couldn't import starfish after installing it, although installing doesn't throw errors really.

I've made either conda environments or python environments in python 3.6, 3.7, 3.8, and 3.10 conda create -n starfish python=3.x then run pip install starfish, and in python of ipython when I run import starfish or any other variation it tries to load it, but then throws the error: ImportError: cannot import name 'register_translation' from 'skimage.feature' then lists where the skimage init.py is.

I've tried to create a new environment and install numpy first and there was no difference. Looks like it's a problem with skimage.

I finally resolved the issue by running pip install scikit-image==0.17.2 after installing starfish, since it seems that in order to be imported, starfish requires the now deprecated function 'register_translation' from 'skimage.feature' as listed in the error message. Could you replace the deprecated function or remove it so that other people don't have as much an issue with installation? Or perhaps add that you need to install that specific version of scikit-image?

ryandikdan commented 2 years ago

Actually, when trying to run the quick tutorial, exactly that function threw an error also. Perhaps this is something you're working out.

berl commented 2 years ago

hi @ryandikdan this has been fixed for a while, but we haven't made a pypi release with the the updates. You should be able to clone the repo and install following the instructions here

berl commented 2 years ago

I'm going to close this, but feel free to reopen if installing from the repo doesn't work.

ryandikdan commented 2 years ago

I did get it to work, although there were some issues. I had to install numpy and napari before running make install-dev. To download it I changed the git:// to https:// compared to the website recommendation. Also for WSL I had to download a server Xming, and update to Windows 11 for WSLg. Otherwise the only issue running the quick start tutorial was that it said the watershed function from scikit-image was being deprecated, but it worked. Hope this helps!

karchern commented 9 months ago

Hi, I ran into the same issue and am also a bit surprised that the release version is so hard to install.

Why isn't the version with the fix released as a version?