Machine learning in computational NMR-aided structural elucidation

Abstract

Structure elucidation is a stage of paramount importance in the discovery of novelcompounds because molecular structure determines their physical, chemical andbiological properties. Computational prediction of spectroscopic data, mainly NMR,has become a widely used tool to help in such tasks due to its increasing easiness andreliability. However, despite the continuous increment in CPU calculation power,classical quantum mechanics simulations still require a lot of effort. Accordingly,simulations of large or conformationally complex molecules are impractical. In thiscontext, a growing number of research groups have explored the capabilities ofmachine learning (ML) algorithms in computational NMR prediction. In parallel,important advances have been made in the development of machine learninginspiredmethods to correlate the experimental and calculated NMR data to facilitatethe structural elucidation process. Here, we have selected some essential papers toreview this research area and propose conclusions and future perspectives for thefield.