This app uses a machine learning model to predict the pKa value of a molecule. The model was trained on a dataset of 24,222 molecules with known pKa values. The model uses the SMILES representation of a molecule as input and outputs the predicted pKa value.
The model was trained using the Random Forest algorithm and achieved an accuracy of 95% on the test set. The model is deployed using the Flask web framework and is hosted on a Heroku server.
The app also includes a molecule editor powered by Ketcher, which allows users to draw a molecule and predict its pKa value.