Simca version 15

The data are presented as the means ± standard errors of the means. Comparisons between two groups accepted the unpaired t-test or Mann-Whitney test after the normality test. Comparisons among multiple groups were analyzed by one-way ANOVA followed by the least significant difference (LSD) post hoc test. Data analyses were performed using statistical product and service solutions (SPSS) 16.0 (IBM, USA), and p < 0.05 indicated a statistically significant difference.
In lipidomic analysis, SIMCA version 15 software (Umetrics, Umeå, Sweden) and MetaboAnalyst software (Version 5.0¹) were used for the PCA, PLS-DA, and OPLS-DA. The volcano plot, box plot, and heatmap were drawn using GraphPad Prism (GraphPad Prism^® Software version 8.0.2 for Windows; La Jolla, CA, United States) and Origin software (Version 2022, OriginLab Inc., USA). The metabolites with variable importance in the projection (VIP), false discovery rate (FDR) adjusted p-value, and absolute log₂Fold change (FC) were used for cluster analysis with Origin software (Version 2022, OriginLab Inc., USA). Violin plots were generated using Hiplot.² Pearson correlations were analyzed by Origin software (Version 2022, OriginLab Inc., USA). In the Pearson correlation and linear mixed-effects model, the significance of each variable was assessed at a level of 0.05.

The MarkerLynx XS^TM version 4.1 (Waters Corporation, Manchester, UK) application manager was used to examine and process the data sets received. The software employed the patented ApexTrack algorithm. The following processing parameters were used: retention time (Rt) ranges from 2–23 min, while the m/z ranged from 50–1200 Da. The Rt and mass windows were each set to 0.20 min and 0.05 Da, respectively. The intensity threshold was set to 100 counts and the mass tolerance to 0.05 Da. For multivariate data analysis (MVDA), the rectified data matrices were then exported to “soft independent modelling of class analogy” (SIMCA-version 15) software (Umetrics, Umea, Sweden). To reduce the dimensionality of the data sets and to study the underlying structures and properties of the data, unsupervised models, such as principal component analysis (PCA) and hierarchical clustering analysis (HCA) were utilised. To compare the two cultivars and find discriminating ions, supervised orthogonal projection to latent structures discriminant analysis (OPLS-DA) was utilised. Validation approaches were then applied to validate the OPLS-DA models and included cross-validated analysis of variance (CV-ANOVA) and receiver operator characteristic (ROC) analysis [33 (link),38 (link),39 (link)].