This repository contains code for the computational screening workflow used in the paper "Exhaustive mapping of the zeolite-template chemical space" by Xie et. al. If you use our code and data, please cite our paper: [TODO insert paper citation]
To install the code, clone the repository and install the dependencies using the following commands:
mamba create -f environment.yml
mamba activate zeobindThe data used in the paper, dataset of hypothetical molecules, as well as predictions across all the IZA framework - hypothetical molecule chemical space, can be found at Materials Data Facility. Download the data and place it in the zeobind/data directory.
We provide a bash script for screening for molecules for a targeted framework using the pre-generated predictions stored inside data/predictions.
Example bash scripts, which are also the ones used in the paper's CHA and ERI case studies, can be found at run_scripts/inference/screen_{cha,eri}.sh. Run
bash run_scripts/inference/screen_cha.shIf you have a new set of molecules you would like to predict on, an example bash script can be found at run_scripts/predict.sh.
bash run_scripts/inference/predict.shTo train models, template bash scripts can be found at run_scripts/training/.
# example
bash run_scripts/training/binary/template.shThe notebooks in the notebooks directory generate metrics and figures reported in the paper.