Machine learning predictions of bond dissociation energies (BDEs)
This tool predicts BDEs for single, noncyclic bonds in neutral organic molecules containing C, H, O, and N atoms. Mean absolute errors are typically less than 1 kcal/mol for most compounds. To use, enter a SMILES string above (or use the drawing tool) and press submit. Reference DFT-calculated BDEs used as training can be displayed for any predicted bond using the neighbor’s link.
Group-contribution predictions of Yield Sooting Index (YSI)
This tool predicts the Yield Sooting Index of a compound as a function of its carbon types. To use, enter a SMILES string above (or use the drawing tool) and press submit. Experimental measurements, when available, are also displayed.
Cetane Number prediction (CNpred)
Cetane number (CN) quantifies the performance and efficiency of diesel fuels, and it is one of the criteria used to design biodiesel candidates. CNpred is the message-passing graph neural network model that predicts the CN of (bio)fuel candidates including hydrocarbons and oxygenates. Only a 2D structure or SMILES string of a molecule is needed to obtain the predicted CN of a given molecule. The code, model, and database used to develop the model are available via GitHub. The website for CN prediction is also available.
Heat of Vaporization prediction (HoVpred)
The heat of vaporization (HoV) is one of the key fuel properties determining particulate matter emission and antiknock characteristics. HoVpred performs a graph attention network-based HoV prediction of fuel candidates at a given temperature by using only 2D structural information of a molecule. The code and model are available via GitHub.
2021 Summer Coding Camp
This GitHub repository contains the materials (Python codes, Jupyter notebooks, and slides) covered by the members of Kim Group and Paton Group in the 2021 Summer Coding Camp. It includes the topics from Python basics to cheminformatics and simple machine learning. The Coding Camp was planned and held to help computational chemistry researchers get the gist of various coding tasks that can potentially be performed in their future research.
