This is the code for the "Predicting Retrosynthetic Reaction using Self-Corrected Transformer Neural Networks" paper found on Journal of Chemical Information and Modeling

To implement our models we were based on OpenNMT-py (v0.4.1).

Create a new conda environment:

conda create -n SCRP_transformer python=3.5
source activate SCRP_transformer
conda install rdkit -c rdkit
conda install future six tqdm pandas

The code was tested for pytorch 0.4.1, to install it go on Pytorch. Select the right operating system and CUDA version and run the command, e.g.:

conda install pytorch=0.4.1 torchvision -c pytorch

Then,

pip install torchtext==0.3.1
pip install -e . 

In the experiments we use an open-source datasets (and train/valid/test splits).

• USPTO-50K dataset data/USPTO-50K
• USPTO-50K-cluster dataset
• Jin’s USPTO dataset

The tokenized datasets can be found on the data/ folder.

For Input file generation, we the preprocess.sh script:

We use a shared vocabulary. The vocab_size and seq_length are chosen to include the whole datasets.

The data has already been preprocessed for training the Self-Corrected Retrosynthetic Reaction Predictor

Model training can be started by running the training.sh script

To achieve the best results with single models, we average the last 10 checkpoints.

To generate the predictions use the translate.py script:

Model testing of greedy search can be started by running the testing_greedy_search.sh script

Model testing of greedy search can be started by running the testing_beam_search.sh script

Run the the score_predictions.sh script to get the top-10 accuracy.

1. use a fully trained model to generate the top ten candidate precursors given a set of target compounds in the training set and validation set.
2. construct a training library that consists of a set of input-output pairs, where the inputs are predicted invalid reactants, and the outputs are the ground truth reactants.
3. get the top-1 candidate produced by the syntax corrector and replace the original invalid smiles.

• run the the self_corrected.sh script to get the top-10 accuracy of the syntax corrector.

If you find this work useful in your research, please consider citing the paper: "Predicting Retrosynthetic Reaction using Self-Corrected Transformer Neural Networks".