Graph Neural Networks for Prediction of Fuel Ignition Quality

This is the web frontend of the graph neural network-based tool for computation of fuel ignition qualities of (oxygenated) hydrocarbons.
The tool takes SMILES strings of a molecule as an input and returns an estimate of the Derived Cetane Number (DCN), Motor Octane Number (MON), and Research Octane Number (RON).
A detailed description of the model is available in our paper. Therein, we also describe the proposed method as well as the training data set in more detail.

Please enter the SMILES string of the molecule:




If you use this tool in your research, please cite our recent work:

@article{Schweidtmann_GNN_Fuel.2020,
  author={Schweidtmann, Artur M. and Rittig, Jan G. and K{\"o}nig, Andrea and Grohe, Martin and Mitsos, Alexander and Dahmen, Manuel},
  title={Graph Neural Networks for Prediction of Fuel Ignition Quality},
  journal={Energy {\&} Fuels},
  pages={11395-11407},
  volume={34},
  number={9},
  issn={0887-0624},
  year={2020},
  doi={https://doi.org/10.1021/acs.energyfuels.0c01533},
}

The table below summarizes the modeling data sets that have been used for the development of the GNN. The table is meant to provide some guidance on the GNNs applicability range.

The machine-learning model is developed in Python using PyTorch and the source code of this tool is freely available at GitLab

Note that the web frontend is currently under development and will be updated soon.