GC–MS characterisation of novel artichoke (Cynara scolymus) pectic-oligosaccharides mixtures by the application of machine learning algorithms and competitive fragmentation modelling

Abstract

Novel artichoke pectic-oligosaccharides (POS) mixtures have been obtained by enzymatic hydrolysis using four commercial enzyme preparations: Glucanex ® 200G, Pentopan ® Mono-BG, Pectinex ® Ultra-Olio and Cellulase from Aspergillus niger. Analysis by HPAEC-PAD showed that Cellulase from A. niger produced the greatest amount of POS (310.6 mg g-1 pectin), while the lowest amount was produced by Pentopan ® Mono-BG (45.7 mg g-1 pectin). To determine structural differences depending on the origin of the enzyme, GC-MS spectra of di-and trisaccharides have been studied employing three machine learning algorithms: multilayer perceptron, random forest and boosted logistic regression. Machine learning models allowed characteristic m/z ions patterns to be established for each enzyme based on their GC-MS spectra with high prediction rates (above 95% on the test set). Possible chemical structures were given for some m/z ions having a decisive influence on these classifications. Finally, it was observed that several ions could be formed from specific POS structures.

Mass Spectrometric Fragmentation of Trimethylsilyl and Related Alkylsilyl Derivatives

“…Other compounds examined are α‐tocopherol (vitamin E) (Liebler et al, ), hydroxypipecolic acids (Kitew & Hughes, ), menthols (Bournot et al, ), tetracyclines (Tsuji & Robertson, ), hydroxynoraporphines (Evans & Vouros, ), nicotinamine (Ripperger, ), flavonoids (Füzfai & Molnár‐Perl, ), bisphenol A (Szyrwińska et al, ), phytoestrogens (Ferrer et al, ), haplamine (Ea et al, ), cucurbic acid (Rontani et al, ), mycolic acids (Kaneda et al, ), N , N ‐dialkyl aminoethane‐2‐ols (Reddy et al, ), tricothecenes (Rodrigues‐Fo et al, ), β‐diketone‐containing plant waxes (Tulloch & Hogge, ), alkoxychlorohydrins (Zenkevich & Drugov, ), abietic acids (Azemard et al, ; Rontani et al, ), pectic oligosaccharides from artichoke (Sabater et al, ), lignans (Boldizsár, et al, ), vitamin B6 (Richter, et al, ) and polysaccharides from the Chinese medicinal herb Anemarrhena asphodeloides Bunge (Xia et al, ). Spectra of these compounds are mainly formed by α‐cleavages or cleavages and/or rearrangements of rings and do not present significant additional information on the general fragmentation of TMS derivatives discussed above.…”

Intestinal anti-inflammatory effects of artichoke pectin and modified pectin fractions in the dextran sulfate sodium model of mice colitis. Artificial neural network modelling of inflammatory markers

“…To determine the monomeric composition of purified APwG and APwA, samples (2% w/v in 0.05 M sodium acetate buffer; pH 5.0) were hydrolysed with 90 U mL −1 of Viscozyme®L preparation and then sugars released were analysed as TMSO by GC-FID, following the method previously described by Sabater et al 16 On the other hand, complete M w distribution profiles of artichoke modified pectins were determined by HPSEC-ELSD. 16,18 The DM was determined by FT-IR. 16,17 All analyses were carried out in duplicate.…”

“…With this aim, machine learning algorithms like artificial neural networks (ANN) have been recently applied to characterize artichoke pectin chains extracted by different methods, 17 and to determine complex structural patterns of pectic oligosaccharides obtained from different pectin sources. 18,19 In addition, these advanced data analysis tools might be employed to develop predictive models of disease course and response to therapy, as well as for characterization of disease heterogeneity and drug development for IBD. 20 Therefore, the aim of this study was to investigate the antiinflammatory effect of artichoke pectin (AP) and modified artichoke pectin fractions (without arabinose, APwA; or galactose, APwG) in a experimental DSS model of colitis in mice.…”

Enzymatic Production and Characterization of Pectic Oligosaccharides Derived from Citrus and Apple Pectins: A GC-MS Study Using Random Forests and Association Rule Learning

“…However, GC-MS can be of great interest because it generates massive amounts of high-dimensional data. 7 To interpret GC-MS-EI results, in silico fragmentation methods have been developed. 13,14 Another important step in MS data analysis involves data modeling to find valuable information about chemical structures.…”

“…ANN have been applied on GC-MS data to classify small molecular structures 19 and compared to RF and boosted logistic regression (BLR) to characterize POS substructures obtained from low methyl-esterified artichoke pectin. 7 However, more studies are needed for determining GC-MS structural differences of POS, for example, POS obtained from an HG type structure and highly methyl-esterified pectin. In addition, in a few studies, association rules between m/z ions for establishing fragmentation patterns that could explain these differences have been calculated.…”