VirMolAnalyte: An AI-Driven <i>In Silico</i> Metabolite Annotation Tool

Metabolites play a crucial role in sustaining biological activities and are also a significant source of new drug development. Nuclear magnetic resonance (NMR) spectroscopy is one of the most important tools for identifying the structures of the metabolites. However, the annotation of NMR spectra relies on costly experimental data. To address this challenge, we present a novel tool, VirMolAnalyte, which enables high-quality structural analysis of ¹³C DEPT NMR spectra without the reliance on experimental data. VirMolAnalyte employs a deep neural network to extract molecular fingerprints from ¹³C DEPT NMR spectra and utilizes a filter-evaluator strategy for multidimensional <i>in silico</i> database searching. When evaluated using DB6123, the filter-evaluator strategy achieved the highest Top 1 accuracy of 94.2%, outperforming the traditional chemical shift search method. In assessments conducted on the largest <i>in silico</i> metabolite database, Metabolites Multidimensional <i>in silico</i> Database, VirMolAnalyte demonstrated an impressive Top 1 accuracy of 90.0% and a Top 5 accuracy of 100.0%. VirMolAnalyte presents an innovative approach for the efficient analysis of metabolites using NMR data, providing significant advantages in new drug discovery and metabolomics research.