The idea of this project is to do exploratory work on the use of graph neural networks (GNNs) to handle several machine learning problems related to molecular chemistry. A first step will be to get familiarized with graph neural networks, namely by using some benchmark datasets (?TUDataset: A collection of benchmark datasets for learning with graphs?, which includes data related to small molecules, but also other applications, from bioinformatics, computer vision, and social networks), as well as with GNN libraries, such as PyTorch Geometric.