HASTESM version 0.9

Written by Samuli Näppi and Tuomo Kalliokoski, Orion Pharma. This software is meant to be run at Orion AWS cloud environment, but you may find it useful nevertheless.

Additional software requirements: You need also Schrödinger Suite (Phase/LigPrep and shape_matching) and slurm.

Hardware: 500 CPU cores for confgen / machine learning prediction, 1 GPU for machine learning training and 40 CPU cores for shape matching.

Anaconda3 is recommended for the installation (create fresh new environment).

conda create -n hastesm-0.9 python=3.11 -y
conda activate hastesm-0.9
mamba install pytorch=2.4.1 pytorch-cuda=12.1 -c pytorch -c nvidia -y
pip install chemprop==2.0.4
mamba install pigz=2.6 -y
pip install git+https://github.com/TuomoKalliokoski/HASTESM

Edit file $CONDA_PREFIX/lib/python3.11/site-packages/hastesm/default_config.txt in the installation to match your environment.

Take a copy example_config.txt and edit it to your needs. Calculation can be started by typing:

hastesm -c my_copy_of_example_config.txt

To update to the latest version:

pip install --upgrade git+https://github.com/TuomoKalliokoski/HASTESM

If you're planning to develop or modify the package, you might want to install it in editable mode:

git clone https://github.com/TuomoKalliokoski/HASTESM
cd repo-name
pip install -e .