LAMAIS: A library-aided approach for efficient 1D 1H NMR qualitative analysis in plant metabolomics

• DOI: 10.1016/j.aca.2024.343100
• Publication Date: 2024-08-14T06:34:36Z
• Authors: Xinyuan Xie; Jiayu Yang; Yuting Lu; Yingting Shi; Jianyang Pan; Haibin Qu
• Keywords: Qualitative analysis

One-dimensional proton nuclear magnetic resonance (1D 1H NMR) spectroscopy is a non-destructive, non-targeted analytical technique providing both qualitative and quantitative insights, particularly beneficial for mixture analysis. However, the qualitative analysis of 1D 1H NMR spectra for mixture samples is laborious and time-consuming, involving extensive database searches and verification experiments like spiking. This process heavily relies on the analyst's expertise, leading to efficiency discrepancies. There is a pressing need for a reliable method to streamline operations and enhance the efficiency of qualitative analysis in complex mixtures.We introduce a library-aided method for spectral profiling, named LAMAIS. This method achieves compound identification through similarity assessment between samples and template data, allowing rapid, automatic compound identification and full-spectrum peak assignment without the need for fitting. LAMAIS correctly identifies over 90 % of components in synthetic mixtures and more than 75 % in experimental mixtures, surpassing other representative methods with a higher F2 score. Our reference library, which currently includes 71 compounds, is tailored to capture the commonality of primary metabolites across diverse plant species. The analysis of real-world samples yielded encouraging results, underscoring LAMAIS's versatility as an auxiliary tool suitable for a variety of botanical sources. For analyst convenience, interactive graphics are utilized as the output format.LAMAIS excels, demonstrating competitiveness and reliability. The approach minimizes repetitive tasks and sample wastage, improving the efficiency of 1D 1H NMR qualitative analysis. Constructing a reference library effectively preserves knowledge, mitigates reliance on human experience, and addresses gaps in the analysis of plant source samples.