.yml file about pesto HOT 3 OPEN

Thanks @rubenalv for the help. I included your instructions in the install instructions of the main README file (commit ba651aa).

I will look into creating a docker image as well.

from pesto.

Hi!

Sorry, I just noticed now that you submitted an issue.

I also noticed a few other people having similar problems with the .yml install.

There are not many dependencies and it shouldn't be too sensitive to the exact version.

The main packages are:

• pytorch (deep learning framework)
• gemmi (for parsing pdb files)
• numpy, scipy, pandas, matplotlib, scikit-learn (the usual tools for numerical analysis)

The optional packages are:

• h5py (storage format for the dataset)
• mdtraj (trajectory files parser)
• tensorboard (visualization toolkit)

The most complicated to install is pytorch. If you want to install it with GPU support, follow the instructions there: https://pytorch.org/

I will push now a new environment file that should be working.

from pesto.

The current .yml file contains a hard-coded path at the end:
prefix: /home/khaos/miniconda3/envs/pesto
Harmless though, see https://stackoverflow.com/questions/41274007/anaconda-export-environment-file.

The current .yml file does not solve the environment for me either, when running
conda env create -f pesto.yml python=3.9

I could create a working conda environment that meets dependencies and runs model.py with:
conda create -n pestocus python=3.9 # python 3.9 to be supported by gemmi
conda activate pestocus
conda install pytorch=2.0.1 pytorch-cuda=11.8 -c pytorch -c nvidia # pytorch with CUDA support
conda install conda-forge::gemmi
conda install conda-forge::tensorboard
conda install numpy scipy pandas matplotlib scikit-learn conda-forge::tqdm
conda install anaconda::h5py

from pesto.