KULL-Centre/_2022_ML-ddG-Blaabjerg

conda install - UnsatisfiableError

Opened this issue · 4 comments

RaSP-log.txt
Hello,
Thank you for sharing this package - it looks really exciting and looking forward to testing it out a bit.

Unfortunately, when I try to install the dependencies with conda, I keep running into UnsatisfiableError problems - the main issue seems to be __glibc though I'm not sure - I've included the full output in the attached file.

Here's the command I'm using:

conda create -n rasp python=3.6 pyyaml=5.3.1 pandas=1.1.4 scipy=1.5.3 numpy=1.17.3 scikit-learn=0.24.0 mpl-scatter-density=0.7 pdbfixer=1.5 pytorch=1.2.0 cudatoolkit=10.0 biopython=1.72 openmm=7.3.1 matplotlib=3.1.1 seaborn=0.11.2 ptitprince=0.2.5 dssp=3.0.0 vaex=4.5.0 -c salilab -c omnia -c conda-forge -c anaconda -c defaults

And what seems to me to be the relevant output:

The following specifications were found to be incompatible with your system:

  - feature:/linux-64::__glibc==2.17=0
  - feature:|@/linux-64::__glibc==2.17=0
  - biopython=1.72 -> libgcc-ng[version='>=7.3.0'] -> __glibc[version='>=2.17']
  - cudatoolkit=10.0 -> libgcc-ng[version='>=10.3.0'] -> __glibc[version='>=2.17']
  - dssp=3.0.0 -> libgcc-ng[version='>=11.2.0'] -> __glibc[version='>=2.17']
  - matplotlib=3.1.1 -> libgcc-ng[version='>=7.3.0'] -> __glibc[version='>=2.17']
  - numpy=1.17.3 -> libgcc-ng[version='>=7.3.0'] -> __glibc[version='>=2.17']
  - pandas=1.1.4 -> libgcc-ng[version='>=7.5.0'] -> __glibc[version='>=2.17']
  - pdbfixer=1.5 -> openmm[version='>=7.2.0'] -> __glibc[version='>=2.17|>=2.17,<3.0.a0']
  - ptitprince=0.2.5 -> libgcc-ng[version='>=9.4.0'] -> __glibc[version='>=2.17']
  - python=3.6 -> libgcc-ng[version='>=9.4.0'] -> __glibc[version='>=2.17']
  - pyyaml=5.3.1 -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - scikit-learn=0.24.0 -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - scipy=1.5.3 -> libgcc-ng[version='>=9.4.0'] -> __glibc[version='>=2.17']

Your installed version is: 2.17

Do you have an environment.yml file available for the environment? I'm wondering if installing that way would make a difference at all. I'll also try installing with miniconda and/or mambaforge as well.

RHEL 7.9, anaconda3/2022.05 (conda 4.13.0)

I switched to mamba, and was able to get everything installed individually except that the vaex and numpy requirements appeared incompatible - forcing numpy=1.17.3 required downgrading to vaex=4.0.0, while forcing vaex=4.5.0 required numpy=1.19.5

Bit of a long shot, but I've run into the exact same error, so I wondered whether you ever managed to get all the dependencies installed and RaSP up and running?

I was able to get it running in the end but required some messing around with the source. If I can find some time I might make a fork with my fixes but a brief description for now:

  1. I've attached my conda/mamba environment. Very important to install BioPython via pip rather than via conda as the version being used is technically incompatible with some of the other packages (I forget which), but it seems to work nonetheless. More recent BioPython versions also do not work unfortunately. You should also be able to ignore plotly/kaleido if I remember correctly - I think I just added those to do some plotting after the fact
    rasp_env.yml.txt
  2. dssp is a problem. I couldn't get any conda version working, so I ended up manually installing dssp 4 (https://github.com/PDB-REDO/dssp) instead and then editing src/pdb_parser_scripts/extract_environments.py to use it.
  3. Actually getting RaSP to run is a bit tricky - I ended up writing my own script based mostly off of the notebook. Again, if I find some time this week I'll make fork this repo with my fixes

@amduffy wondering if you're able to make a fork with said fixes? would appreciate the help!