Chromatof software

All analyses were conducted for 3 biological replicates. For metabolomics analysis, GC–TOF-MS and UHPLC–LTQ-MS/MS raw data were converted to NetCDF format (*.cdf) using LECO Chroma TOF™ software (version 4.44, LECO Corp., St. Joseph, MI, USA) and Xcalibur software (version 2.00, Thermo Fisher Scientific), respectively. After conversion, peak alignment was processed using MetAlign software (version 041012, Wageningen, Netherlands, http://www.metalign.nl), and the resulting data were exported to an Excel spreadsheet. Multivariate statistical analysis was performed using SIMCA-P+ 12.0 software (Umetrics; Umea, Sweden). The significantly altered metabolites under mineral deficiencies in P. frutescens were selected based on variable importance projection (VIP) values, and significance was tested by analysis of variance (ANOVA) and t-test using PASW Statistics 18 software (SPSS, Inc., Chicago, IL, USA). Selected metabolites were tentatively identified by comparing mass spectra, retention times, mass fragment patterns, UV absorbances, and elemental compositions derived from UHPLC–LTQ-MS/MS and UPLC–Q-TOF-MS analyses considering standard compounds, an in-house library, and published references. The heatmap was constructed by MeV software (http://www.tm4.org/).

Preprocessed data obtained from the Leco Chroma TOF software (ver. 4.50; Leco) were processed using BinBase, an in-house database [16 (link)–18 (link)]. The peak intensity of the identified metabolites was normalized by the median of the sum of the peak intensities of all identified metabolites in each sample. STATISTICA (ver. 7.1; StatSoft, Tulsa, OK) was used for univariate analyses such as principal component analysis (PCA), nonparametric Wilcoxon-Mann-Whitney test, and breakdown and the one-way analysis of variance (ANOVA) [18 (link),19 (link)]. Orthogonal partial least square-discriminative analysis (OPLS-DA) and seven-fold internal cross validation was performed using SIMCA-P+ (version 12.0; Umetric AB, Umea, Sweden). MedCalc software (Broekstraat, Mariakerke, Belgium) was used to obtain a receiver operating characteristic (ROC) curve for further evaluation of the diagnostic properties of identified metabolites.

Data preprocessing and statistical analysis were performed according to a previous study^{11 (link)}. Raw data collected from GC-TOF-MS were converted into CDF (NetCDF) files with a LECO CHROMA TOF software (version 4.44; LECO Corp.). Peak detection, retention time correction, and alignment were then processed with an online METALIGN software package (http://www.metalign.nl). After that, final data were transformed into Excel files. Data were then arranged on a three-dimensional matrix, including information on the peak, peak area, and sample name. Multivariate statistical analysis was performed with SIMCA-P⁺ (version 12.0, Umetrics; Umea, Sweden). Unsupervised principal component analysis (PCA) was conducted to investigate the general aggregate state and trends in different groups among all samples. A model of supervised orthogonal partial least squares discriminant analysis (OPLS-DA) was performed to maximize metabolic change and determine significantly alternated metabolites in different groups. Variables were separated based on variable importance in the projection (VIP) value. Finally, the statistical difference between control and treatment groups was assessed using independent (unpaired) samples t-tests for unequal variances in SAS 9.4 (SAS Institute Inc., Cary, NC, USA). Differences observed between means were considered significant at p < 0.05.

The results of behavioural tests and test indexes were analysed and processed by SPSS 26.0, and the data are expressed as mean ± standard deviation ( $\overline{\mathrm{x}}$ ± $\mathrm{S}$ ). The comparison between multiple groups was processed by single-factor analysis of variance, and the comparison between the two samples was analysed by LSD test. Metabonomics data were imported into SIMCA after the original data are processed by LECO chroma TOF software for multidimensional statistical analysis. The specific standards of potential differential metabolites are variable importance in the projection (VIP) > 1 and p < 0.05 (according to t-test). Substance was identified by the NIST database and the database of Shanghai Jiaotong University Analysis and Testing Centre. The involved metabolic pathways were investigated by the MetaboAnalyst platform.

After the GC-TOF-MS and UHPLC-LTQ-ESI-IT-MS/MS analyses, raw data files were converted into a computable document format (*.cdf) using LECO Chroma TOF software (Version 4.4, LECO Corp.) and using the thermo file converter program in Thermo Xcalibur software (version 2.1, Thermo Fisher Scientific). The acquired NetCDF format (*.cdf) files were processed to determine retention time, baseline correction, peak detection, and alignment using the MetAlign software package [35 ]. After alignment, the resulting data file (*.csv) including the corrected peak retention times, peak areas, and corresponding mass (m/z) data was exported to Microsoft Excel (Microsoft, Redmond, WA, USA) for further analysis. Multivariate statistical analysis using SIMCA-P+ 12.0 software (version 12.0, Umerics, Umea, Sweden) was performed to compare metabolite differences between commercial gochujang samples by unsupervised principal component analysis (PCA), supervised partial least squares discriminant analysis (PLS-DA), and orthogonal PLS-DA (OPLS-DA). Significant differences (p-value < 0.05) between selected metabolites, their relative contents, and the values from the physicochemical characteristics assays were evaluated by one-way analysis of variance using SPSS 18.0 (SPSS Inc., Chicago, IL, USA). The differences between metabolites were visualized by heat map analysis using MEV software [36 ].

The raw data obtained from GC-TOF-MS were converted into the Net CDF format using LECO Chroma TOF software (LECO Corporation). The raw UHPLC-LTQ-Orbitrap-MS/MS data were converted to the Net CDF format using Xcalibur software (Thermo Fisher Scientific). The converted CDF files were preprocessed using MetAlign software for alignment based on peak detection and retention time correction. SIMCA-P+ software (Umetrics, Umeå, Sweden) was used for multivariate statistical analysis. Significantly different metabolites were selected by variable importance in projection (VIP > 1.0) values based on partial least squares discriminant analysis (PLS-DA) score plots and a significance test (p-value < 0.05) by analysis of variance (ANOVA) using the STATISTICA 7 software (StatSoft, Tulsa, OK, USA). The selected compounds were tentatively identified by comparing their retention times, mass fragment patterns, and elemental compositions with those of standard compounds, published data, and/or public databases such as the National Institute of Standards and Technology (NIST) Library (version 2.0, FairCom, Gaithersburg, MD, USA), Wiley8, and the Human Metabolome Database (HMDB; http://www.hmdb.ca/).

Metabolites separated by GC × GC-TOF-MS were identified using LECO Chroma TOF software by reference to online and local databases, including the NIST 2014 mass spectral database (Scientific Instrument Services, Inc. NJ, USA) and LECO/Fiehn Metabolite mass spectral library (Version 1.00, LECO Corp.). Fiehn retention index values were calculated using LECO Chroma TOF software with reference to the fatty acid methyl ester (C6-C24)^{22 (link)}. Mass spectral matching was manually supervised with a match threshold of >650 (maximum 1000). Peak areas for each metabolite were based on selected quantification masses.
UPLC-IMS-QTOF-MS metabolites identification was performed in UNIFI Scientific Information System software (Waters Corp.) and based comparison of accurate mass, retention time, MS² fragments and CCS values with online reference databases including Respect (http://spectra.psc.riken.jp/), Metlin (https://metlin.scripps.edu/), HMDB (http://www.hmdb.ca/), Lipidmap (http://www.lipidmaps.org/), in-house databases based on commercial standards and theoretical MS² tags, and bibliographies. The CCS value acceptable error was <5%; with MS tolerance of 5 p.p.m., and MS/MS tolerance of <10 mDa, at least one major fragment was found.