AI-augmented R-group exploration in medicinal chemistry

This repo accompanies the preprint "AI-augmented R-group exploration in medicinal chemistry".

Setting up the environment

Create a new conda environment and install rdkit: conda create -c conda-forge -n <your-env-name> rdkit

Activate your environment and install the rest of the requirements: conda activate <your-env-name> pip install -r requirements.txt

How to run

cd data/drd2

Train the model

../../r_fit.py

Predict the activity of test compounds

../../r_predict.py

Peform the Free-Wilson analysis, preferably by training a new model on the entire dataset ref.smi+test.smi

../../r_FW.py

Get help

../../r_fit.py -h

Notes

The core is defined in SMARTS strings. The kekular structure of benzene C1=C-C=C-C=C1 is not valid.