Abstract

OCAMPO-MORALES, Blanca Nayelli; HERNANDEZ-MONTES, Arturo  and  HERBERT-PUCHETA, José Enrique. Near-Infrared Untargeted Metabolomics with Unsupervised and Supervised Multivariate Statistical Analysis of Fatty Acid Profiles in Cheeses. J. Mex. Chem. Soc [online]. 2025, vol.69, n.3, pp.636-651.  Epub Feb 20, 2026. ISSN 1870-249X.  https://doi.org/10.29356/jmcs.v69i3.2212.

The present work describes a workflow for unsupervised Principal Component (PCA) and supervised Partial Least Squares Discriminant (PLS-DA) multivariate statistical analysis (MSA), to analyze Near Infrared (NIR) data matrixes of cheeses from diverse types and geographical origins, with respect to their NIR saturated fatty acid profile. The data set include (A) acquired NIR absorbance spectra, (B) post-processed first derivative NIR spans and (C) post-processed first derivative frequency-selected NIR spans, within a wavelength range of 12500-3600 cm-1. NIR data inputs were adapted for the first time into a format suitable for the stream-lined metabolomics data analysis “MetaboAnalyst”, by converting spectrophotometer raw data format, into a JCAMP-DX IUPAC standard format family for spectral data exchange, in turn transformed into an editable comma-separated values (.csv) format, suitable for metabolomics studies with MetaboAnalyst. The discriminant regions for the first NIR data matrix were five. For the second matrix, discriminant wave-number regions were reduced to three: 10000 to 8000 cm-1 (-CH- overtone), 6000 to 5000 cm-1(-C=O- overtone) and 5000 to 4000 cm-1 (-CH- band). Finally, for the third NIR matrix, refined discriminant regions were taken: 9700 to 8265 (-CH- overtone), 6661 to 4655 cm-1 (-C=O- overtone) and from 4327 to 4000 cm-1 (-CH- band). The PLS-DA model obtained from the first derivative frequency-selected near-infrared spans data matrix showed the best score-plot classification between dairy samples and saturated fatty acid standards. Present results intend to introduce an approach for untargeted and qualitative NIR based metabolomics within a platform with more than 300,000 users to date.

Keywords : Near infrared spectroscopy; NIR based metabolomics; cheeses; untargeted metabolomics; saturated fatty acid (SFA).

        · abstract in Spanish     · text in English     · English ( pdf )