This repository contains the code to download the OCELOT chromophore dataset and use the pretrained models to make predictions. You can find the online implementation at here
More details on model in our paper Electronic, Redox, and Optical Property Prediction of Organic P-Conjugated Molecules through a Hierarchy of Machine Learning Approaches
eval.ipynb - make predictions with SMILES
MPNN_evidential - the MPNN model with evidential deep learning
dataset.ipynb - download the OCELOT chrmomophore v1 dataset and transform to pandas DataFrame
mlp_features and normalize_feats.csv - generate the features for model input
first_gen_models.ipynb - Training of first generation model
If you use the dataset or any trained models in your work, please cite the following article-
Bhat, V.; Sornberger, P.; Pokuri, B. S. S.; Duke, R.; Ganapathysubramanian, B.; Risko, C. Electronic, Redox, and Optical Property Prediction of Organic P-Conjugated Molecules through a Hierarchy of Machine Learning Approaches. Chemical Science 2022. https://doi.org/10.1039/d2sc04676h.