A trained multitask constraint message passing neural networks for QM atomic/bond property predictions as described in the paper Regio-Selectivity Prediction with a Machine-Learned Reaction Representation and On-the-Fly Quantum Mechanical Descriptors.
QM descriptors under B3LYP/def2svp level of theory that can be predicted with this model:
- Hirshfeld partial charge
- Neucleuphilic Fukui indices
- Electrophilic Fukui indices
- NMR shielding constants
- Bond lengths
- Bond orders
Documentation: Documentation of qmdesc is available at https://qmdesc.readthedocs.io/en/latest/index.html.
- RDKit
For all installations, we recommend using conda to get the necessary rdkit dependency:
conda install -c rdkit rdkit
pip install qmdesc
Or from envrioment.yml
conda create --name qmdesc --file environment.yml