NMR-Onion - a transparent multi-model based 1D NMR deconvolution algorithm.

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Computational Chemistry

Background:

• Nuclear Magnetic Resonance (NMR) spectroscopy is crucial for molecular structure elucidation.
• Automatic deconvolution of 1D NMR spectra is challenging due to complex line shapes and overlapping signals.
• Existing methods may struggle with asymmetric non-Lorentzian line shapes and can lead to excessive peak detection.

Purpose of the Study:

• To introduce NMR-Onion, an open-source, computationally efficient algorithm for automated 1D NMR spectral deconvolution.
• To develop a tool capable of handling asymmetric non-Lorentzian line shapes.
• To improve the precision and efficiency of NMR spectral analysis.

Main Methods:

• Development of an algorithm using Python and PyTorch.
• Implementation of two innovative time-domain models for spectral deconvolution.
• Inclusion of resolution-enhanced peak detection and digital filtering capabilities.
• Integration of three built-in statistical models with automatic selection via the Bayesian Information Criterion (BIC).
• Assessment of result repeatability through post-modeling uncertainty evaluation.

Main Results:

• NMR-Onion provides computationally efficient and automatic deconvolution of 1D NMR spectra.
• The algorithm successfully handles asymmetric non-Lorentzian line shapes.
• Precise peak prediction and efficient computation are achieved through core components.
• Automatic selection of statistical models and evaluation of uncertainty enhance reliability.
• Excessive peak detection is minimized, leading to cleaner spectral analysis.

Conclusions:

• NMR-Onion offers a robust and efficient solution for automated 1D NMR spectral deconvolution.
• The algorithm's ability to handle complex line shapes and minimize false positives improves data interpretation.
• NMR-Onion represents a valuable open-source tool for researchers in chemistry and related fields.