SMART‐Miner: A convolutional neural network‐based metabolite identification from ¹H‐¹³C HSQC spectra

Abstract

The identification of metabolites from complex biofluids and extracts of tissues is an essential process for understanding metabolic profiles. Nuclear magnetic resonance (NMR) spectroscopy is widely used in metabolomics studies for identification and quantification of metabolites. However, the accurate identification of individual metabolites is still a challenging process with higher peak intensity or similar chemical shifts from different metabolites. In this study, we applied a convolutional neural network (CNN) to 1H‐13C HSQC NMR spectra to achieve accurate peak identification in complex mixtures. The results reveal that the neural network was successfully trained on metabolite identification from these 2D NMR spectra and achieved very good performance compared with other NMR‐based metabolomic tools.

Analytical and Structural Tools of Lipid Hydroperoxides: Present State and Future Perspectives

“… Incorporation of the numerous literature 1 H and 13 C-NMR chemical shift and coupling constant data of lipid hydroperoxides into open-source NMR (web) databases. These data combined with prediction software will greatly facilitate structural information from 1 H, 13 C, 1 H- 1 H COSY, 1 H- 1 H TOCSY, 1 H- 13 C HSQC, and 1 H- 13 C HMBC NMR data [ 194 , 195 , 196 , 197 ]. DFT calculations of 1 H-, and 13 C- chemical shifts and J couplings can contribute significantly to the unequivocal resonance assignment, identification of cis / trans geometric isomers, and diastereomeric pairs of complex hydroperoxides and solvent effects [ 142 , 143 , 144 , 145 , 146 , 147 , 148 , 149 , 150 , 151 , 152 , 153 , 154 , 155 , 156 , 198 , 199 ].…”

“…Incorporation of the numerous literature 1 H and 13 C-NMR chemical shift and coupling constant data of lipid hydroperoxides into open-source NMR (web) databases. These data combined with prediction software will greatly facilitate structural information from 1 H, 13 C, 1 H- 1 H COSY, 1 H- 1 H TOCSY, 1 H- 13 C HSQC, and 1 H- 13 C HMBC NMR data [ 194 , 195 , 196 , 197 ].…”

NMR Metabolomics Methods for Investigating Disease

“…For example, using computational approaches, an automated metabolite identification and assignment approach using the 1 H– 1 H TOCSY experimental data was introduced, and it was shown to assign peaks accurately . In another approach, a convolutional neural network (CNN) was applied to 2D 1 H– 13 C HSQC NMR spectra . The neural network was trained for metabolite identification from the 2D NMR spectra and achieved performance comparable to the conventional approaches.…”

“…26 In another approach, a convolutional neural network (CNN) was applied to 2D 1 H− 13 C HSQC NMR spectra. 27 The neural network was trained for metabolite identification from the 2D NMR spectra and achieved performance comparable to the conventional approaches. Machine learning offers opportunities for automated analysis of NMR spectra 28 and potentially represents a new dimension to the NMR-based metabolomics field.…”

Convolutional Neural Network-Based Compound Fingerprint Prediction for Metabolite Annotation

“…For instance, Zhang et al trained DeepSpectra, a CNN module for pattern recognition from raw near infrared spectral data [ 15 ]. Kim et al applied CNN and developed SMART-Miner for identifying 2D NMR peaks from a mixture sample for metabolite identification [ 16 ]. Fedorova et al applied and found one-dimensional CNN predicts the most accurate retention time in reversed-phase liquid chromatography [ 17 ].…”