Synergy of NMR, DFT, and artificial intelligence methods used in identifying diastereomers

What is it about?

Computational NMR spectroscopy, which provides reliable coupling constants nJXY (n = 1-3) and chemical shifts dX based on routine DFT/GIAO predictions, combined with advanced probabilistic methods (including the Goodman’s CP3 parameter or numerous options of type DP4 and beyond), and artificial neural networks play a significant role in the validation or revision of structures of diastereomeric species, especially natural compounds, in all cases where a single-crystal X-ray structural analysis cannot be performed. This Digest article (review paper) critically discusses the above topics and other closely related issues, such as the Boltzmann statistic for flexible multiconformer molecules and linear regression of calculated versus experimental NMR chemical shifts dX (where X = H, C, and N).

Featured Image

Why is it important?

In recent years, it has become apparent that many previous structural assignments for different diastereoisomeric systems, especially those involving complex natural products, carried out solely based on classical degradation and/or chemical derivatization and analysis of 1D/2D NMR experimental data sets for solutions, were erroneous. The use of various modern DFT computational methods, described here extensively, at the molecular structure determination stage should result in no further misassignments in the literature.

Perspectives