Code repository for Multilabel Classification Models for the Prediction of Cross-Coupling Reaction Conditions, Michael R. Maser*, Alexander Y. Cui*, Serim Ryou*, Travis J. DeLano, Yisong Yue, Sarah E. Reisman J. Chem. Inf. Model. 2021, 61, 156.

Including:

Chainer implementation of Graph Neural Networks for the Prediction of Substrate-Specific Organic Reaction Conditions, Serim Ryou*, Michael R. Maser*, Alexander Y. Cui*, Travis J. DeLano, Yisong Yue, Sarah E. Reisman, ICML 2020 Graph Representation Learning and Beyond (GRL+) Workshop. arXiv:2007.04275

1. Install chainer-chemistry
2. Download the dataset by following the instruction from Reaxys®
3. This code supports the label dictionary of suzuki, CN coupling, Negishi and PKR datasets.
4. Training command (with gpu)

python train.py -m <METHOD> -e <NUM_EPOCHS> -o <OUTPUT_DIR> -g 0 --data-name <One from suzuki, CN, Negishi or PKR>

Example:

python train.py -m relgcn -e 50 -o relgcn_output -g 0 --data-name suzuki

3. Testing command (with gpu)

python predict.py -m <METHOD> -i <DIR_WITH_MODEL> -g 0 --load-modelname <FILEPATH_TO_MODEL> --data-name <One from suzuki, CN, Negishi or PKR>

Example:

python predict.py -m relgcn -i relgcn_output -g 0 --load-modelname relgcn_output/model_epoch-1 --data-name suzuki

4. Modify the path to the result directory in convert_to_evaluation_format.ipynb and generate json files.