cultivarium / MicrobeMod

A toolkit for exploring prokaryotic methylation and base modifications in nanopore sequencing
MIT License
34 stars 1 forks source link

Conda package #2

Open Ge0rges opened 8 months ago

Ge0rges commented 8 months ago

Hello,

I was wondering if there are any plans to create a conda package to facilitate install?

alexcritschristoph commented 8 months ago

Conda would be nice... will try to see if it can be done in the future. Note that most of the external dependencies can be installed with conda:

conda install -c bioconda prodigal hmmer blast -c nanoporetech modkit

However, I don't think it would be possible to package STREME in conda which would be a limitation. We might be able to remove this dependency in the future though.

alexweisberg commented 8 months ago

I believe the MEME suite is on bioconda as well (as package meme), which should include STREME already.

aforestsomewhere commented 8 months ago

Just tagging onto to confirm I managed to get MicrobeMod running with the bioconda distro of meme (which was great as the compilation on the HPC was proving to be a real pain in the hoop).

Briefly:

conda create -n microbemod_env mamba
mamba install -c bioconda prodigal hmmer blast meme
mamba install -c nanoporetech modkit #had to install this separately as it was not resolving dependencies out of the box
#DL cath-resolve hits binary
chmod u+x cath-resolve-hits
mv cath-resolve-hits ../conda/microbemod_env/bin
export PATH=/path/to/cath-resolve-hits:$PATH

There was a slight issue with it not finding the path to the site packages in my conda virtual environment, so I had to manually add it with sys.append() to get it to pass the tests.

Thanks for the great software, trying it out on some fresh ONT libraries now!

Ge0rges commented 8 months ago

When I try that with pandas 2.1.2 py310hcc13569_0 conda-forge in my condo environment I get an error running pip install .: ERROR: Could not install packages due to an OSError: [Errno 2] No such file or directory: '/Accounts/gkanaan/.conda/envs/methylate/lib/python3.10/site-packages/pandas-2.1.1.dist-info/METADATA'

I fixed this just by changing the version. Any idea why that would happen given the requirements aren't that strict?

alyssah7 commented 7 months ago

Thanks for developing this tool -- I am excited to use it! I have been struggling to get cath-resolve-hits downloaded. It seems as though the binaries are not compatible with my HPC (CentOS rather than Ubuntu), and the build from source is running into all sorts of dependency errors as the software versions still on the cath-tools GitHub are outdated. I feel a bit stuck in that never-ending dependency hairball.

Have you all see this issue on the development side? Do you have any recommendations for how to move forward?

alexcritschristoph commented 7 months ago

I've not run into that issue, although I'm on ubuntu. ...

Are you able to try installing it this way through conda instead? https://anaconda.org/bioconda/cath-tools

I am exploring the possibilities of a conda package and a Docker for MicrobeMod that might help - hope to have an update soon.

alexweisberg commented 7 months ago

Our cluster runs Centos 7 and the download of cath-resolve-hits worked ok for us. I made the conda environment then downloaded just cath-resolve-hits from the github and put it in the bin folder of the conda environment. I had to make it executable with chmod +x cath-resolve-hits before it would run.

alyssah7 commented 7 months ago

@alexcritschristoph None of their docs pointed to a conda download, so I just never tried -- that worked! Thanks for the assist!