AttenGpKa

AttenGpKa is a graph neural network for predicting pKa of different molecules in various solvents. If you use resources of this project, please cite:

1. H. An, X. Liu, W. Cai, X. Shao. AttenGpKa: A Universal Predictor of Solvation Acidity Using Graph Neural Network and Molecular Topology. Journal of Chemical Information and Modeling, 2024. DOI: 10.1021/acs.jcim.4c00449

The training data, which is collected from the iBond database, can be found in the Supporting Information of our published paper (https://doi.org/10.1021/acs.jcim.4c00449).

If you use this data in your work, in addition to this work, please also cite:

2. J.-D. Yang, X.-S. Xue, P. Ji, X. Li, J.-P. Cheng, Internet Bond-energy Databank (pKa and BDE): iBonD Home Page. http://ibond.nankai.edu.cn.

1 environment requirements

The required packages and their versions are included in the requirements.txt file. Run the following commands to build your environment:

conda create -y -n AttenGpKa python==3.8.11
conda activate AttenGpKa
pip install -r requirements.txt
conda install -y ipython

2 usage

We open-sourced all the data, codes, and models used in this work. Additionally, we have developed a user-friendly software for Windows OS, which allows anyone to easily use our model.

3 model training and predicting

If users want to train their own model with our provided training data:

python AttenGpka.py -m train -s ./test.h5 -f 0

-m train or --mode train represents the training mode -s ./test.h5 or --saveweights ./test.h5 is the path of the trained model to be saved
-f 0 or --fold 0 is the way to split training and test dataset

If users want to predict pKa with the available model:

python AttenGpka.py -m predict -l ./model.h5 -d ./test_data.csv

-l /trained_model/model.h5 or --loadweights /trained_model/model.h5 is the path of weights to be loaded
-d ./test_data.csv or --data ./test_data.csv is the path of data to be predicted

4 Notification of commercial use

Commercialization of this product is prohibited without our permission.

5 Citing this work

Our published data is collected from the iBond database. If you use this data in your work, please cite:

J.-D. Yang, X.-S. Xue, P. Ji, X. Li, J.-P. Cheng, Internet Bond-energy Databank (pKa and BDE): iBonD Home Page. http://ibond.nankai.edu.cn or  http://ibond.chem.tsinghua.edu.cn.