The recent biomolecular studies tend to involve combinations of different methods and approaches that allow analyzing organisms at the genomic and proteomic levels, as well as at the level of metabolomics. However, in order to justify the use of the metabolomics techniques in plant breeding, it is important to comprehensively analyze a broad range of species and varieties. In this study, we evaluated the contents of low-molecular-weight substances in seeds of different rapeseed cultivars by the gas chromatography–mass spectrometry (GC–MS) technique. For every metabolomic profile, we estimated 168 target substances, and 52 of them were unambiguously identified. These compounds included amino acids, organic and fatty acids, tocopherols, and phytosterols. In order to keep the data assay within the context of multivariate statistics, we used principal component analysis (PCA), partial least square discriminant analysis (PLS-DA), and partial least square regression (PLS-R). The subsequent analysis revealed a significant difference between the metabolomic profiles of the investigated rapeseed cultivars, with the primary role of the amino acids and organic acids. Noticeably, the PCA and PLS-DA models showed 65% of the explained variance and, according to the Venetian blinds’ cross-validation test, 91.67% accuracy. Thus, we demonstrated the effectiveness of the metabolomic approach for the varietal identification of seeds. This strategy can be further improved with a continuously updated database of the metabolomic profiles of different species and cultivars. Application of the PLS-DA method will make it possible to compare metabolites of unknown seed samples with the existing metabolomic profiles and, subsequently, identify new seed samples.
- Brassica napus L
- gas chromatography
- mass spectrometry
- principal component analysis
- projection to latent structures discriminant analysis