Structure-Based Predictions of ¹H NMR Chemical Shifts Using Feed-Forward Neural Networks

Abstract

Feed-forward neural networks were trained for the general prediction of 1H NMR chemical shifts of CH(n) protons in organic compounds in CDCl3. The training set consisted of 744 1H NMR chemical shifts from 120 molecular structures. The method was optimized in terms of selected proton descriptors (selection of variables), the number of hidden neurons, and integration of different networks in ensembles. Predictions were obtained for an independent test set of 952 cases with a mean average error of 0.29 ppm (0.20 ppm for 90% of the cases). The results were significantly better than those obtained with counterpropagation neural networks.

Fast and accurate algorithm for the simulation of NMR spectra of large spin systems

“…This approach relies on the accuracy of the prediction of the NMR parameters and on the accuracy of the simulation of the spectrum. Several algorithms available for the prediction of chemical shifts and scalar coupling constants are fast and reliable [5][6][7][8][9][10][11][12][13]. Unfortunately, the simulation of the spectrum using spin dynamics scales exponentially with the number of atoms, and thereby making the available algorithms unsuitable for large systems [14][15][16][17][18][19][20][21][22][23][24].…”

Synthesis of a Polyconjugated Polymer by Aldol Condensation of 2,5-Diformylfuran and Acetone

“…1 H NMR spectra were recorded using a Bruker Avance III 600 MHz spectrometer (Krasnoyarsk Regional Research Equipment Center of SB RAS). Theoretical simulation of NMR spectra based on compound structure was accomplished using the open web-service SPINUS-WEB [8][9][10][11][12]. Electron spin resonance (hereinafter ESR) spectra were recorded using a Bruker Elexsys…”

Improving the efficiency of branch-and-bound complete-search NMR assignment using the symmetry of molecules and spectra

“…Indeed, it is the most common procedure for structure validation by NMR, thus playing a key role in structure elucidation and validation by NMR spectroscopy, which in turn is a core component of new compound discovery/synthesis and related fields. Furthermore, most of the research in the field of NMR chemical shift prediction [5][6][7][8][9][10][11][12][13][14] and automatic elucidation [15][16][17][18][19][20][21] depends on repositories of well-assigned NMR data. Although manual assignment by an expert is the most widely used and most reliable method, the already big and continuously growing amount of information produced nowadays demands computational tools for assisting this task.…”