Prediction of 15 N chemical shifts by machine learning.

We demonstrate the potential for machine learning systems to predict three-dimensional (3D)-relevant NMR properties beyond traditional 1 H- and 13 C-based data, with comparable accuracy to density functional theory (DFT) (but orders of magnitude faster). Predictions of DFT-calculated 15 N chemical shifts for 3D molecular structures can be achieved using a machine learning system-IMPRESSION (Intelligent Machine PREdiction of Shift and Scalar information Of Nuclei), with an accuracy of 6.12-ppm mean absolute error (∼1% of the δ15 N chemical shift range) and an error of less than 20 ppm for 95% of the chemical shifts. It provides less accurate raw predictions of experimental chemical shifts, due to the limited size and chemical space diversity of the training dataset used in its creation, coupled with the limitations of the underlying DFT methodology in reproducing experiment.

A community-powered search of machine learning strategy space to find NMR property prediction models.

The rise of machine learning (ML) has created an explosion in the potential strategies for using data to make scientific predictions. For physical scientists wishing to apply ML strategies to a particular domain, it can be difficult to assess in advance what strategy to adopt within a vast space of possibilities. Here we outline the results of an online community-powered effort to swarm search the space of ML strategies and develop algorithms for predicting atomic-pairwise nuclear magnetic resonance (NMR) properties in molecules. Using an open-source dataset, we worked with Kaggle to design and host a 3-month competition which received 47,800 ML model predictions from 2,700 teams in 84 countries. Within 3 weeks, the Kaggle community produced models with comparable accuracy to our best previously published 'in-house' efforts. A meta-ensemble model constructed as a linear combination of the top predictions has a prediction accuracy which exceeds that of any individual model, 7-19x better than our previous state-of-the-art. The results highlight the potential of transformer architectures for predicting quantum mechanical (QM) molecular properties.

IMPRESSION - prediction of NMR parameters for 3-dimensional chemical structures using machine learning with near quantum chemical accuracy.

The IMPRESSION (Intelligent Machine PREdiction of Shift and Scalar information Of Nuclei) machine learning system provides an efficient and accurate method for the prediction of NMR parameters from 3-dimensional molecular structures. Here we demonstrate that machine learning predictions of NMR parameters, trained on quantum chemical computed values, can be as accurate as, but computationally much more efficient (tens of milliseconds per molecular structure) than, quantum chemical calculations (hours/days per molecular structure) starting from the same 3-dimensional structure. Training the machine learning system on quantum chemical predictions, rather than experimental data, circumvents the need for the existence of large, structurally diverse, error-free experimental databases and makes IMPRESSION applicable to solving 3-dimensional problems such as molecular conformation and stereoisomerism.