| DS Description
|
Data Processing<br />
<br />
N … Data Processing<br />
<br />
Non-processed MS data from GC-TOF-MS analysis were exported in NetCDF format, which was generated by the chromatography processing and mass spectral deconvolution software (Leco ChromaTOF version 2.32; LECO, MI, USA), to MATLAB 7.0 and MATLAB2011b (Mathworks, MA, USA), where all data pretreatment procedures, such as smoothing, alignment, time-window setting and hierarchical multivariate curve resolution (H-MCR), were carried out. The resolved mass spectra were matched against the reference mass spectra by using the NIST mass spectral search program from the NIST/EPA/NIH mass spectral library (Data Version: NIST 08, Software Version 2.0), Terpenoid library, Identification of Essential Oil Components by Gas Chromatography/Mass Spectrometry (the 4th Edition) commercial library and VocBinbase. Our custom software was also used for peak annotation written in Java. The metabolites were defined as annotated metabolites by comparison with RI values calculated by alkane standard mixtures (C8–20) and mass spectra from these three libraries. Peaks that showed a match factor of ≥800 in the mass spectral library were selected for annotation. Data were normalized by dividing each peak area by the peak area of n-tridecane and metabolite identifiers were organized using MetMask.<br />
<br /><br />
Data Analysis<br />
<br />
The two data matrices (observations: biological replicates, variables: annotated peaks) were used for statistical analyses, respectively (see Supplementary Tables S1–3). Multivariate analysis was performed using the SIMCA-P+13.0 software (Umetrics AB, Umeå, Sweden), with log10 transformation and unit variance scaling. Metabolite profile data were log2 transformed and then statistically analyzed using the LIMMA package, which includes false discovery rate (FDR) correction for multiple testing (Benjamini and Hochberg, 1995) in the R environment for statistical computing (version 2.14.2). Analysis of variance testing of cross-validated predictive residuals (CV-ANOVA) was used as the significance test for assessing the reliability of O2PLS-DA models (significant, p cv < 0.01). S-DA models (significant, p cv < 0.01).
|
