This repository holds the code for the publication: XXX.
The dependencies for running these scripts include:
- rdkit
- numpy
- pandas
- matplotlib
The code here should be sufficient for reproducing the main paper figures. To reproduce all figures, please follow the following steps. Numbered scripts within a directory should be run in numerical order.
-
First compute Emin from scratch for the QM9 dataset, by running the scripts in numbered order within the directory Computing Emin for QM9. You will first need to download files within the links found in Computing Emin for QM9/data/raw/README.md
-
Next, compute other synthesizabaility scores for the QM9 dataset by running all scripts within Computing Other Scores. Each script will take the output from step 1, calculate the corresponding score, and overwrite the file from step 1 with the additional score.
- install ASKCOS from here: https://github.com/ASKCOS/askcos-core/tree/main and specify the installation path at the top of sa_sc_syba_score_parallel.py
- install RA score from here: https://github.com/reymond-group/RAscore
- install SYBA score from here: https://github.com/lich-uct/syba
-
Use the Mordred descriptor calculator in Computing Mordred Features for QM9 to compute general descriptors for ML. This will take the file from step 1 (now with all calculated other scores from step 2), calculate Mordred descriptors, and write a new file to this step's directory.
- install Mordred from here: https://github.com/mordred-descriptor/mordred
-
Open Reproduce paper figures and run each script in any order.