2.7. Data Processing

¹H-NMR (600 MHz) spectra for statistical and quantitative analyses were first processed using the TopSpin software (Bruker Corp.). Free induction decays were multiplied by a 0.3 Hz exponential multiplication function prior to Fourier transformation; the tetramethylsilane (TMS) signal was calibrated at δ 0.00, and only a zero-order phase correction was allowed.

For statistical analysis of spectra, the binning of 0.04 ppm was applied to spectral data using MestreNova software, and spectra were transformed into a data matrix. The MetaboAnalyst 3.0 platform (http://www.metaboanalyst.ca/faces/home.xhtml accessed on 1 December 2020) was used for principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA). No data filtering, no sample normalization, and Pareto scaling (mean-centered and divided by the square root of the standard deviation of each variable) were used in data preprocessing. Leave-one-out cross-validation (LOOCV) was applied in PLS-DA. The accuracy, variable importance in projection (VIP) and clustering results shown as heatmaps (distance measure using euclidean, clustering algorithm using ward.D, view options only group aver-ages of top 15 PLS-DA VIP) were also assessed.

For quantitative purposes, specific ¹H-NMR signals were manually integrated, and the concentrations of omega-3 (L–linolenic acid) and omega-6 (Ln–linoleic acid) type fatty acids were calculated following the method previously reported by others [26,27]. The concentrations of fatty acids were expressed in molar percentages according to Equations (1) and (2).

in which A_omega-3 and A_omega-6 are the areas of the bis-allylic proton peaks for omega-3 and omega-6 fatty acids, respectively, and A_G is the area of the proton peaks of glyceryl groups; L refers to omega-3, linolenic acid, and Ln refers to omega-6, linoleic acid.

For statistical analysis of the ratio of omega-3/omega-6, data analyses were performed using the GraphPad Prism version 7.0 for Windows (GraphPad Software Corp., La Jolla, CA, USA). Standard parametric (Student’s t-test and one-way Analysis of variance, ANOVA) tests were applied accordingly to variables type and distribution, with post-hoc test Dunnett’s for multiple comparisons. All distribution was checked using a Shapiro–Wilk test. All results are expressed as the value of mean ± standard deviation (SD). The significance threshold was considered at p ≤ 0.05.