From NMR to AI: Do We Need 1H NMR Experimental Spectra to Obtain High-Quality logD Prediction Models?
Open Access
- 5 March 2025
- journal article
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (6) , 2924-2939
Abstract
This study presents a novel approach to 1H NMR-based machine learning (ML) models for predicting logD using computer-generated 1H NMR spectra. Building on our previous work, which integrated experimental 1H NMR data, this study addresses key limitations associated with experimental measurements, such as sample stability, solvent variability, and extensive processing, by replacing them with fully computational workflows. Benchmarking across various density functional theory (DFT) functionals and basis sets highlighted their limitations, with DFT-based models showing relatively high RMSE values (average CHI logD of 1.12, lowest at 0.96) and extensive computational demands, limiting their usefulness for large-scale predictions. In contrast, models trained on predicted 1H NMR spectra by NMRshiftDB2 and JEOL JASON achieved RMSE values as low as 0.76, compared to 0.88 for experimental spectra. Further analysis revealed that mixing experimental and predicted spectra did not enhance accuracy, underscoring the advantage of homogeneous datasets. Validation with external datasets confirmed the robustness of our models, showing comparable performance to commercial software like Instant JChem, thus underscoring the reliability of the proposed computational workflow. Additionally, using normalized RMSE (NRMSE) proved essential for consistent model evaluation across datasets with varying data scales. By eliminating the need for experimental input, this workflow offers a widely accessible, computationally efficient pipeline, setting a new standard for ML-driven chemical property predictions without experimental data constraints.Keywords
This publication has 57 references indexed in Scilit:
- An Introduction to Statistical Learning2023
- Machine learning in drug design: Use of artificial intelligence to explore the chemical structure–biological activity relationshipWIREs Computational Molecular Science, 2021
- A nifty collaborative analysis to predicting a novel tool (DRFLLS) for missing values estimationSoft Computing, 2019
- Facilitating quality control for spectra assignments of small organic molecules: nmrshiftdb2 – a free in‐house NMR database with integrated LIMS for academic service laboratoriesMagnetic Resonance in Chemistry, 2015
- 1H and 13C NMR scaling factors for the calculation of chemical shifts in commonly used solvents using density functional theoryJournal of Computational Chemistry, 2014
- Four‐component relativistic chemical shift calculations of halogenated organic compoundsJournal of Physical Organic Chemistry, 2013
- Quantitative prediction of 13C NMR chemical shifts in solvent using PCM‐ONIOM method and optimally selected wave functionConcepts in Magnetic Resonance Part A, 2013
- Effect of the damping function in dispersion corrected density functional theoryJournal of Computational Chemistry, 2011
- The M06 suite of density functionals for main group thermochemistry, thermochemical kinetics, noncovalent interactions, excited states, and transition elements: two new functionals and systematic testing of four M06-class functionals and 12 other functionalsTheoretical Chemistry Accounts, 2007
- The Solvent Dimethyl SulfoxideAngewandte Chemie International Edition, 1967