Geometric Deep Learning For Molecular Crystal Structure Prediction

New machine learning strategies have been developed and tested for accelerating molecular crystal structure ranking and crystal property prediction using tools from geometric deep learning on molecular graphs.

Mark Tuckerman (New York University)

New machine learning strategies have been developed and tested 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, models were trained for density prediction and stability ranking which are accurate, fast to evaluate, and applicable to molecules of widely varying size and composition. The team’s 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. The crystal ranking tool, MolXtalNet-S, correctly discriminates experimental samples from synthetically generated fakes and is further validated through analysis of the submissions to Cambridge Structural Database Blind Tests. These 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.

Designing Materials to Revolutionize and Engineer our Future (DMREF)