Improved Carbohydrate Structure Generalization Scheme for ¹H and ¹³C NMR Simulations

The improved Carbohydrate Structure Generalization Scheme has been developed for the simulation of ¹³C and ¹H NMR spectra of oligo- and polysaccharides and their derivatives, including those containing noncarbohydrate constituents found in natural glycans. Besides adding the ¹H NMR calculations, we improved the accuracy and performance of prediction and optimized the mathematical model of the precision estimation. This new approach outperformed other methods of chemical shift simulation, including database-driven, neural net-based, and purely empirical methods and quantum-mechanical calculations at high theory levels. It can process structures with rarely occurring and noncarbohydrate constituents unsupported by the other methods. The algorithm is transparent to users and allows tracking used reference NMR data to original publications. It was implemented in the Glycan-Optimized Dual Empirical Spectrum Simulation (GODESS) web service, which is freely available at the platform of the Carbohydrate Structure Database (CSDB) project (http://csdb.glycoscience.ru).