DESPERATE: A Python library for processing and denoising NMR spectra

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Computational Chemistry

Background:

• Nuclear Magnetic Resonance (NMR) spectroscopy suffers from low signal-to-noise ratio (SNR) due to inherent sensitivity limitations and random thermal noise.
• Current methods to improve SNR, like signal averaging and apodization, have drawbacks including long experimental times or loss of spectral resolution.
• Singular-value decomposition (SVD) based denoising methods are effective but computationally expensive for large NMR datasets.

Purpose of the Study:

• To introduce a novel wavelet transform (WT) based routine for efficient and robust denoising of 1D and 2D NMR spectra.
• To compare the performance of WT denoising against established SVD-based methods (Cadzow, PCA) in terms of SNR enhancement and spectral uniformity.
• To provide a fast and accessible solution for improving NMR spectral quality.

Main Methods:

• Implementation of a new wavelet transform (WT) routine for NMR spectral denoising.
• Application of WT denoising to simulated and experimental 1D and 2D NMR datasets.
• Comparative analysis of WT denoising against SVD-based methods (Cadzow, PCA) using metrics like SNR enhancement and spectral uniformity.

Main Results:

• Wavelet transform (WT) denoising demonstrated comparable or superior SNR enhancement and spectral uniformity compared to SVD-based methods.
• WT denoising achieved significantly faster processing times, in some cases by several orders of magnitude.
• The developed WT routine is effective for both 1D and 2D NMR spectra.

Conclusions:

• Wavelet transform (WT) offers a fast, robust, and effective alternative for denoising NMR spectra, overcoming limitations of traditional methods.
• The WT approach provides significant computational advantages over SVD-based techniques for large NMR datasets.
• All routines are available in the free and open-source Python library DESPERATE, promoting accessibility and further research.