Geometric Deep Learning for Molecular Crystal Structure Prediction

We develop and test new machine learning strategies for accelerating molecular crystal structure ranking and crystal property prediction using tools from geometric deep learning on molecular graphs. Leveraging developments in graph-based learning and the availability of large molecular crystal data sets, we train models for density prediction and stability ranking which are accurate, fast to evaluate, and applicable to molecules of widely varying size and composition. Our density prediction model, MolXtalNet-D, achieves state-of-the-art performance, with lower than 2% mean absolute error on a large and diverse test data set. Our crystal ranking tool, MolXtalNet-S, correctly discriminates experimental samples from synthetically generated fakes and is further validated through analysis of the submissions to the Cambridge Structural Database Blind Tests 5 and 6. Our new tools are computationally cheap and flexible enough to be deployed within an existing crystal structure prediction pipeline both to reduce the search space and score/filter crystal structure candidates.

[1]  Pascal Friederich,et al.  Graph neural networks for materials science and chemistry , 2022, Communications Materials.

[2]  A. White,et al.  Symmetric Molecular Dynamics , 2022, Journal of chemical theory and computation.

[3]  K. Fountoulakis,et al.  Graph Attention Retrospective , 2022, ArXiv.

[4]  S. Pozdnyakov,et al.  Incompleteness of graph neural networks for points clouds in three dimensions , 2022, Mach. Learn. Sci. Technol..

[5]  Nicholas D. Lane,et al.  Do We Need Anisotropic Graph Neural Networks? , 2021, ICLR.

[6]  S. Botti,et al.  Crystal graph attention networks for the prediction of stable materials , 2021, Science advances.

[7]  Shuiwang Ji,et al.  Spherical Message Passing for 3D Graph Networks , 2021, ArXiv.

[8]  Michael Gastegger,et al.  Equivariant message passing for the prediction of tensorial properties and molecular spectra , 2021, ICML.

[9]  N. Marom,et al.  Crystal structure prediction of energetic materials and a twisted arene with Genarris and GAtor , 2021, CrystEngComm.

[10]  N. Marom,et al.  Machine Learned Model for Solid Form Volume Estimation Based on Packing-Accessible Surface and Molecular Topological Fragments. , 2020, The journal of physical chemistry. A.

[11]  Johannes T. Margraf,et al.  Fast and Uncertainty-Aware Directional Message Passing for Non-Equilibrium Molecules , 2020, ArXiv.

[12]  Benjamin Kurt Miller,et al.  Relevance of Rotationally Equivariant Convolutions for Predicting Molecular Properties , 2020, ArXiv.

[13]  Dominique Beaini,et al.  Principal Neighbourhood Aggregation for Graph Nets , 2020, NeurIPS.

[14]  Stephan Günnemann,et al.  Directional Message Passing for Molecular Graphs , 2020, ICLR.

[15]  Chris Wolverton,et al.  Developing an improved crystal graph convolutional neural network framework for accelerated materials discovery , 2019, 1906.05267.

[16]  Mojtaba Haghighatlari,et al.  A deep neural network model for packing density predictions and its application in the study of 1.5 million organic molecules , 2019, Chemical science.

[17]  Markus Meuwly,et al.  PhysNet: A Neural Network for Predicting Energies, Forces, Dipole Moments, and Partial Charges. , 2019, Journal of chemical theory and computation.

[18]  Chi Chen,et al.  Graph Networks as a Universal Machine Learning Framework for Molecules and Crystals , 2018, Chemistry of Materials.

[19]  Jure Leskovec,et al.  How Powerful are Graph Neural Networks? , 2018, ICLR.

[20]  Padmini Rajagopalan,et al.  MT-CGCNN: Integrating Crystal Graph Convolutional Neural Network with Multitask Learning for Material Property Prediction , 2018, ArXiv.

[21]  K. Reuter,et al.  Genarris: Random generation of molecular crystal structures and fast screening with a Harris approximation. , 2018, The Journal of chemical physics.

[22]  Álvaro Vázquez-Mayagoitia,et al.  GAtor: A First-Principles Genetic Algorithm for Molecular Crystal Structure Prediction. , 2018, Journal of chemical theory and computation.

[23]  Li Li,et al.  Tensor Field Networks: Rotation- and Translation-Equivariant Neural Networks for 3D Point Clouds , 2018, ArXiv.

[24]  Takashi Miyake,et al.  Crystal structure prediction accelerated by Bayesian optimization , 2018 .

[25]  K-R Müller,et al.  SchNet - A deep learning architecture for molecules and materials. , 2017, The Journal of chemical physics.

[26]  Jeffrey C Grossman,et al.  Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties. , 2017, Physical review letters.

[27]  Qiang Zhu,et al.  DDT Polymorphism and the Lethality of Crystal Forms. , 2017, Angewandte Chemie.

[28]  Claire S. Adjiman,et al.  Report on the sixth blind test of organic crystal structure prediction methods , 2016, Acta crystallographica Section B, Structural science, crystal engineering and materials.

[29]  Jesse G. McDaniel,et al.  Next-Generation Force Fields from Symmetry-Adapted Perturbation Theory. , 2016, Annual review of physical chemistry.

[30]  Jian Lv,et al.  Crystal structure prediction via particle-swarm optimization , 2010, 1008.3601.

[31]  Peter T. A. Galek,et al.  Knowledge-based model of hydrogen-bonding propensity in organic crystals. , 2007, Acta crystallographica. Section B, Structural science.

[32]  Junmei Wang,et al.  Development and testing of a general amber force field , 2004, J. Comput. Chem..

[33]  Frank H. Allen,et al.  Cambridge Structural Database , 2002 .

[34]  Thomas Lengauer,et al.  Derivation of a scoring function for crystal structure prediction. , 2001, Acta crystallographica. Section A, Foundations of crystallography.

[35]  W. Motherwell.  Crystal Structure Prediction and the Cambridge Structural Database , 2001 .

[36]  Jan Kroon,et al.  Upack program package for crystal structure prediction: Force fields and crystal structure generation for small carbohydrate molecules , 1999, J. Comput. Chem..

[37]  K. Hamacher,et al.  Stochastic Tunneling Approach for Global Minimization of Complex Potential Energy Landscapes , 1999, physics/9903008.

[38]  Thomas Lengauer,et al.  Crystal Structure Prediction based on Statistical Potentials , 1998 .

[39]  Robin Taylor,et al.  IsoStar: A library of information about nonbonded interactions , 1997, J. Comput. Aided Mol. Des..