RESUMEN
Plant samples are potential sources of physiologically active secondary metabolites and their classification is an extremely important task in traditional medicine and other fields of research. In the production of herbal drugs, different plant parts of the same or related species can serve as adulterants for primary plant material. The use of highly informative and relatively easily accessible tools, such as liquid chromatography and low-resolution mass spectrometry, helps to solve these tasks by means of fingerprint analysis. In this study, to reveal specific plant part features for 20 species from one family (Apiaceae), and to preserve the maximum information content, two approaches are suggested. In both cases, minimal raw data pretreatment, including rescaling of time and m/z axes and cutting off some uninformative regions, was applied. For the support vector machine (SVM) method, tensor unfolding was required, while neural networks (NNs) were able to work directly with squared heatmaps as input data. Moreover, five data augmentation variants are proposed, to overcome the typical problem of a lack of data. As a result, a comparable F1-score close to 0.75 was achieved by SVM and two employed NN architectures. Eight marker compounds belonging to chlorophylls, lipids, and coumarin apio-glucosides were tentatively identified as characteristic of their corresponding sample groups: roots, stems, leaves, and fruits. The proposed approaches are simple, information-saving and can be applied to a broad type of tasks in metabolomics.
RESUMEN
The task of multipurpose analysis of biological samples and identification of individual compounds in them is actual for many organizations in various fields; the results of such analyses can affect lives. The most frequently used, most accurate, and highly sensitive method used for this kind of analysis is the combination of gas/liquid chromatography and high-resolution mass spectrometry. However, in some areas, it is necessary to increase the reliability of compound identification. In this paper, we present a method that combines the reaction of oxygen isotope exchange with mass spectrometry; the method allows to increase the reliability of identification of individual compounds. Oxygen isotope exchange reaction is a "selective" one, which means that not all oxygen present in the molecule can exchange, but only in certain functional groups. Thus, by the number of isotope exchanges that have occurred in this molecule, the right structural formula might be more accurately chosen. The method was tested both on pure pharmaceutical substances and on real human urine samples. In both cases, the effectiveness of the method was shown: the number of expected exchanges in known substances coincided with the experimental one, and from several possible structures of unknown substances, the correct one was chosen based on the number of isotope exchanges.
Asunto(s)
Oxígeno , Cromatografía de Gases y Espectrometría de Masas , Humanos , Espectrometría de Masas/métodos , Isótopos de Oxígeno , Reproducibilidad de los ResultadosRESUMEN
The combination of Liquid Chromatography and Mass Spectrometry (LC-MS) is commonly used to determine and characterize biologically active compounds because of its high resolution and sensitivity. In this work we explore the interpretation of LC-MS data using multivariate statistical analysis algorithms to extract useful chemical information and identify clusters of similar samples. Samples of leaves from 19 plants belonging to the Apiaceae family were analyzed in unified LC conditions by high- and low-resolution mass spectrometry in a wide range scan mode. LC-MS data preprocessing was performed followed by statistical analysis using tensor decomposition in the form of Parallel Factor Analysis (PARAFAC); matrix factorization following tensor unfolding with principal component analysis (PCA), independent component analysis (ICA), non-negative matrix factorization (NMF); or unsupervised feature selection (UFS). The optimal number of components for each of these methods were found and results were compared using four different metrics: silhouette score, Davies-Bouldin index, computational time, number of noisy components. It was found that PCA, ICA and UFS give the best results across the majority of the criteria for both low- and high-resolution data. An algorithm for biomarker signal selection is suggested and 23 potential chemotaxonomic markers were tentatively identified using MS2 data. Dendrograms constructed by the methods were compared to the molecular phylogenic tree by calculating pixel-wise mean square error (MSE). Therefore, the suggested approach can support chemotaxonomic studies and yield valuable chemical information for biomarker discovery.