SMART deep learning tools to accelerate the characterization of natural product structures from their NMR data

Abstract

Diverse terrestrial and marine organisms produce biologically active natural products, many of which have inspired some of humanity's most effective medicines. A critical step in developing natural product-based therapeutics is the complete elucidation of molecular structure, a process that integrates multiple spectroscopic techniques, with nuclear magnetic resonance (NMR) spectroscopy playing a central role. However, interpreting NMR data requires significant expertise and access to costly instrumentation, posing challenges to efficient structural characterization. To address this, we have developed two complementary artificial intelligence tools, SMART 2.1 and DeepSAT, which assist in identifying structurally related molecules based on a compound's 1H–13C HSQC NMR spectrum. This paper presents step by step instructions for using these tools to accelerate the structure elucidation of novel natural products. © 2025 Elsevier B.V., All rights reserved.