forfreedomgit/localtransform

Predicting Organic Reactivity Based on Generalized Reaction Templates

LocalTransform

Implementation of organic reactivity prediction with LocalTransform developed by prof. Yousung Jung group at KAIST (contact: [email protected]).

Contents

• Developer
• OS Requirements
• Python Dependencies
• Installation Guide
• Reproduce the results
• Demo and human benchmark results
• Citation
• License

Developer

Shuan Chen ([email protected])

OS Requirements

This repository has been tested on both Linux and Windows operating systems.

Python Dependencies

• Python (version >= 3.6)
• Numpy (version >= 1.16.4)
• PyTorch (version >= 1.0.0)
• RDKit (version >= 2019)
• DGL (version >= 0.5.2)
• DGLLife (version >= 0.2.6)

Installation Guide

Create a virtual environment to run the code of LocalTransform.
Make sure to install pytorch with the cuda version that fits your device.
This process usually takes few munites to complete.

git clone https://github.com/kaist-amsg/LocalTransform.git
cd LocalTransform
conda create -c conda-forge -n rdenv  python=3.6 -y
conda activate rdenv
conda install pytorch cudatoolkit=11.3 -c pytorch -y
conda install -c conda-forge rdkit -y
conda install -c dglteam dgl-cuda11.3
pip install dgllife

Reproduce the results

[1] Download the raw data of USPTO-480k dataset

Download the data from https://github.com/wengong-jin/nips17-rexgen/blob/master/USPTO/ and move the data to ./data/USPTO_480k/.

[2] Data preprocessing

A two-step data preprocessing is needed to train the LocalTransform model.

1) Local reaction template derivation

First go to the data processing folder

cd preprocessing

and extract the reaction templates.

python Extract_from_train_data.py

This will give you four files, including (1) real_templates.csv (reaction templates for real bonds) (2) virtual_templates.csv (reaction templates for imaginary bonds) (3) template_infos.csv (including the hydrogen change, charge change and action information)

2) Assign the derived templates to raw data

By running

python Run_preprocessing.py

You can get four preprocessed files, including (1) preprocessed_train.csv (2) preprocessed_valid.csv (3) preprocessed_test.csv (4) labeled_data.csv

[3] Train LocalTransform model

Go to the main scripts folder

cd ../scripts

and run the following to train the model with reagent seperated or not (default: False)

python Train.py -sep True

The trained model will be saved at LocalTransform/models/LocalTransform_sep.pth

[4] Test LocalTransform model

To use the model to test on test set, simply run

python Test.py -sep True

to get the raw prediction file saved at LocalTransform/outputs/raw_prediction/LocalTransform_sep.txt
Finally you can get the reactants of each prediciton by decoding the raw prediction file

python Decode_predictions.py -sep True

The decoded reactants will be saved at LocalTransform/outputs/decoded_prediction/LocalTransform_sep.txt

[5] Exact match accuracy calculation

By using

python Calculate_topk_accuracy.py -m sep

the top-k accuracy will be calculated from the files generated at step [4]

Demo and human benchmark results

See Synthesis.ipynb for running instructions and expected output. Human benchmark results is also shown at the end of the notebook.
It takes approximately one second to predict a product from a set of given reactants.

Citation

A Generalized Template-Based Graph Neural Network for Accurate Organic Reactivity Prediction, Nat Mach Intell 2022

License

This project is covered under the Apache 2.0 License.