Chemprop is an open-source implementation of a directed message passing neural network (D-MPNN) that has been demonstrated to be successful in predicting a variety of molecular properties, including solvation properties, optical properties, infrared spectra, and toxicity. The framework has also been used to predict reaction properties and to identify new antibiotics. Chemprop can be used via a command-line interface or within a Python script or Jupyter notebook. In this tutorial, we will discuss the D-MPNN algorithm and several successful use cases, followed by a demonstration of some of Chemprop’s core functionalities (e.g. training and testing models for single- and multiple-molecule properties and reactions, transfer learning, and estimating uncertainty).

Kevin Greenman is a Ph.D. candidate in chemical engineering and computation at MIT working under the supervision of Profs. Rafael Gómez-Bombarelli and William Green. During his undergraduate studies at the University of Michigan, he studied the structural, thermodynamic, and optical properties of nitride semiconductors, and he developed a nanoHUB tool for research and education on computational catalysis during a summer REU at Purdue University. His current research interests are in using the combination of atomistic simulations and machine learning with experimental collaboration for the discovery and design of new molecules and materials.

Network for Computational Nanotechnology (NCN)

Researchers should cite this work as follows:

• Kevin Greenman (2022), "Message-Passing Neural Networks for Molecular Property Prediction Using Chemprop," https://nanohub.org/resources/36082.

  BibTex | EndNote

1. chemistry
2. Chemprop
3. hands on
4. Jupyter notebooks
5. machine learning
6. materials science
7. message passing neural networks (MPNN)
8. Molecular Properties
9. NCN Group - Machine Learning Hands-on
10. neural networks
11. python