A new PARAFAC‐based algorithm for HPLC–MS data treatment: herbal extracts identification

Abstract

Introduction: Role of highly informative high-performance liquid chromatography mass spectrometry (HPLC-MS) methods in quality control is increasing. Complex herbal products and formulations can simultaneously contain extracts from different plants. Therefore, due to the leads to lack of commercial standards it is important to develop novel approaches for comprehensive treatment of big datasets. Objective: The aim of this study is to create a straightforward and information-saving algorithm for the identification of plants extracts in commercial products. Material and methods: In total, 34 samples, including Glycyrrhiza glabra and Panax ginseng dried roots; and Abrus precatorius dried leaves, their double and triple mixtures and flavoured oolong tea samples were analysed by HPLC-MS and combined in a three-dimensional dataset (retention time-mass-to-charge ratio (m/z)-samples). This dataset was subjected to smoothing and denoising techniques and further decomposed using parallel factor analysis (PARAFAC). Results: Samples were divided into eight clusters; loading matrices were interpreted and the presence of the most characteristic triterpene glycoside groups was demonstrated and supported by the characteristic chromatogram approach. The occurrence of Abrus precatorius and G. glabra additives in flavoured tea was confirmed. Conclusion: Developed HPLC-MS-PARAFAC method is potentially reliable and an efficient tool for handling untreated experimental data and its future development may lead to more comprehensive evaluation of chemical composition and quality control of food additives and other complex mixtures.

Two Different Strategies Based on Three‐Way Analysis Models of UV Spectroscopic and Spectrochromatographic Measurements to Quantify the Acidity Constant of Acetylsalicylic Acid

“…with the use of a wide range of analytical techniques such as spectrofluorimetry, HPLC, CE, which can measure the analytical signal of a sample in more than one dimension. [30][31][32][33][34][35] It is also possible to obtain higher order data arrays with more conventional instruments such as UV/VIS by incorporating a new dimension such as reaction time, pH change, and sample.…”

Dissection of the Active Ingredients and Potential Mechanism of Han-Shi-Yu-Fei-Decoction in Treating COVID-19 Based on In Vivo Substances Profiling and Clinical Symptom-Guided Network Pharmacology

“…As 108 compounds were found from database 1.0, the retention times of those compounds were added in the library list, and therefore generated the three-dimensional data set (retention time-m/z-compounds) in Personal Compound Database Library (PCDL), which certainly was the database 2.0. − Normally, the method criterion for retention time was 0.25 min, and the average mass error of adduct and isotopes was 10 ppm. The rat serum was analyzed by the same UHPLC–Q-TOF-MS method, and therefore the absorbed prototype compounds were screened out by quick-matching the peaks in the database 2.0 using the Agilent MassHunter Qualitative Analysis 10.0 software.…”

Hopomics: Humulus lupulus Brewing Cultivars Classification Based on LC-MS Profiling and Nested Feature Selection

“…Dimensionality reduction techniques can be performed in two ways: feature selection by statistical tests or machine learning methods to eliminate statistically non-informative variables and feature extraction by principal component analysis (PCA) or other algorithms in order to transform the whole feature space into new variables. PCA and other similar data representation techniques [ 31 ] are also often used in the first steps of LC-MS experimental data treatment for exploratory purposes, such as quality assurance [ 23 ]. In some cases, it is even possible to cluster sample dots for group labels on a score plot by visual analysis of the first two or three principal components spaces [ 32 , 33 ].…”