Computational methods for the prediction of NMR spectra are well established. Our group has previously developed the DP4 parameter, which provides a confidence level for the match between calculated and experimental NMR spectra. This method has been widely used to help assign the structures of many stereochemically-rich molecules. More recently this method has been fully automated and is available as open-source package PyDP4. This software integrates conformational search, DFT calculations and the statistical analysis and permits rapid and convenient computational elucidation of structure and relative stereochemistry.<br/>We have tested various combinations of functionals and basis sets, and mPW1PW91 is optimal for relative stereochemistry elucidation. This can be further improved by calculating conformer energies at M06-2X level. We have also systematically tested the effect of number and quality of conformer geometries on the DP4 process accuracy and provide data to demonstrate how best to distribute computer resources between conformation analysis and DFT sophistication. The underlying statistical methods of DP4 have been revised. We demonstrate that models that are sum of 2-3 Gaussians offer the best balance.<br/>The importance of a solvent model in NMR calculations is well known, but implicit solvent models do not capture H-bonding interactions with the solute. We have explored the effects of explicit solvent models on NMR prediction and structure elucidation accuracy, and provide recommendations of the optimal choices.<br/>One throughput-limiting step in structure elucidation is the NMR description preparation. To remove this bottleneck, we have developed processes for the automated interpretation of raw NMR data and incorporated them in to the PyDP4 workflow. DP4 analysis of medium-sized molecules used to take several days of operator time, including manual analysis of the NMR data. Now, FID to a conclusion of even greater confidence takes less than an hour of active effort.<br/>In summary, improved strategies for resource allocation, more sophisticated statistical models, and better analyses of solvent effects and operator intervention-free treatment of FIDs straight from the spectrometer have all been integrated into the PyDP4 workflow. The resulting process is faster, more accurate and less labour-intensive, enabling the analysis of more complex systems in larger numbers.

Advances in computational NMR structure elucidation by DP4: Automation, optimisation and raw data analysis

256th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.