Semi-quantitative chemometric models for characterization of mixtures of sugars using infrared spectral data
Anna Luiza B. Brito,Inês F. Cardoso,Luís P. Viegas,Rui Fausto
DOI: https://doi.org/10.1016/j.saa.2024.125225
IF: 4.831
2024-10-05
Spectrochimica Acta Part A Molecular and Biomolecular Spectroscopy
Abstract:Sugars (saccharides) are sweet-tasting carbohydrates that are abundant in foods and play very important roles in living organisms, particularly as sources and stores of energy, and as structural elements in cellular membranes. They are desirable therapeutic targets, as they participate in multiple metabolic processes as fundamental elements. However, the physicochemical characterization of sugars is a challenging task, mostly due to the structural similarity shared by the large diversity of compounds of this family. The need for fast, accurate enough, and cost-effective analytical methods for these substances is of extreme relevance, in particular because of the recently increasing importance of carbohydrates in Medicine and food industry. With this in view, this work focused on the development of chemometric models for semi-quantitative analysis of samples of different types of sugars (glucose, galactose, mannitol, sorbose and fructose) using infrared spectra as data, as an example of application of a novel approach, where the Principal Component Analysis (PCA) score plots are used to estimate the composition (weight-%) of the mixtures of the sugars. In these plots, polygonal geometric shapes emerge in the vectorial space of the most significant principal components, that allow grouping different types of samples on the vertices, edges, faces and interior of the polygons according to the composition of the samples. This approach was applied successfully to mixtures of up to 5 sugars and shown to appropriately extract the compositional information from the hyper-redundant complex spectral data. Thought the method has been applied here to a specific problem, it shall be considered as a general procedure for the semi-quantitative analysis of other types of mixtures and applicable to other types of data reflecting their composition. In fact, the methodology appears as an efficient tool to solve three main general problems: (i) use hyper-redundant (in variables) data, as spectral information, directly and with minimum pre-treatment, to evaluate semi-quantitatively the composition of mixtures; (ii) do this for systems which produce data that can be considered rather similar; and (iii) do it for a number of substances present in the mixtures that might be greater than that usually considered in chemistry, which in general is limited to 3 components. In addition, this work also demonstrates that, similarly to the developed analysis based on the PCA score plots, the Multivariate Curve Resolution with Alternating Least Squares (MCR-ALS) chemometric method can also be used successfully for the qualitative (when used without any previous knowledge of the components present in the samples) or semi-quantitative (when the pure components spectral profiles are provided as references) analyses of mixtures of (at least) up to 5 distinct sugars.
spectroscopy