Prediction of Molecular Packing Motifs in Organic Crystals by Neural Graph Fingerprints
Material search is a significant step for discovery of novel materials with desirable characteristics, which normally requires exhaustive experimental and computational efforts. For a more efficient material search, neural networks and other machine learning techniques have recently been applied to materials science in expectation of their potentials in data-driven estimation and prediction. In this study, we aim to predict molecular packing motifs of organic crystals from descriptors of single molecules using machine learning techniques. First, we identify the molecular packing motifs for molecular crystals based on geometric conditions. Then, we represent the information on single molecules using the neural graph fingerprints which are trainable descriptors unlike conventional untrainable ones. In numerical experiments, we show that the molecular packing motifs are better predicted by using the neural graph fingerprints than the other tested untrainable descriptors. Moreover, we demonstrate that the key fragment of molecules in crystal motif formation can be found from the trained neural graph fingerprints. Our approach is promising for crystal structure prediction.
KeywordsCrystal structure prediction Machine learning Supervised learning Neural graph fingerprints
- 13.Duvenaud, D.K., et al.: Convolutional networks on graphs for learning molecular fingerprints. In: Advances in Neural Information Processing Systems, pp. 2224–2232 (2015)Google Scholar
- 15.Glen, R.C., Bender, A., Arnby, C.H., Carlsson, L., Boyer, S., Smith, J.: Circular fingerprints: flexible molecular descriptors with applications from physical chemistry to adme. IDrugs 9(3), 199 (2006)Google Scholar
- 17.Gilmer, J., Schoenholz, S.S., Riley, P.F., Vinyals, O., Dahl, G.E.: Neural message passing for quantum chemistry. arXiv preprint arXiv:1704.01212 (2017)
- 19.Kingma, D.P., Ba, J.: Adam: a method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014)
- 20.Snoek, J., Larochelle, H., Adams, R.P.: Practical Bayesian optimization of machine learning algorithms. In: Advances in Neural Information Processing Systems, pp. 2951–2959 (2012)Google Scholar