School of Korean Medicine, Dongshin University, Naju 58245, Korea.
Department of Pharmacy, College of Pharmacy, Mokpo National University, Mokpo 58554, Korea.
Molecules. 2019 Oct 24;24(21):3837. doi: 10.3390/molecules24213837.
The purpose of this study was to analyze metabolic differences of ginseng berries according to cultivation age and ripening stage using gas chromatography-mass spectrometry (GC-MS)-based metabolomics method. Ginseng berries were harvested every week during five different ripening stages of three-year-old and four-year-old ginseng. Using identified metabolites, a random forest machine learning approach was applied to obtain predictive models for the classification of cultivation age or ripening stage. Principal component analysis (PCA) score plot showed a clear separation by ripening stage, indicating that continuous metabolic changes occurred until the fifth ripening stage. Three-year-old ginseng berries had higher levels of valine, glutamic acid, and tryptophan, but lower levels of lactic acid and galactose than four-year-old ginseng berries at fully ripened stage. Metabolic pathways affected by different cultivation age were involved in amino acid metabolism pathways. A random forest machine learning approach extracted some important metabolites for predicting cultivation age or ripening stage with low error rate. This study demonstrates that different cultivation ages or ripening stages of ginseng berry can be successfully discriminated using a GC-MS-based metabolomic approach together with random forest analysis.
本研究旨在使用基于气相色谱-质谱联用（GC-MS）的代谢组学方法，分析人参果的代谢差异，依据的是其种植年龄和成熟阶段。在三年生和四年生人参的五个不同成熟阶段，人参果每周收获一次。利用鉴定出的代谢物，采用随机森林机器学习方法，获得用于分类种植年龄或成熟阶段的预测模型。主成分分析（PCA）得分图通过成熟阶段清晰地区分，表明在第五个成熟阶段之前持续发生了代谢变化。在完全成熟阶段，三年生人参果的缬氨酸、谷氨酸和色氨酸水平较高，但乳酸和半乳糖水平较低，而四年生人参果则相反。受不同种植年龄影响的代谢途径涉及到氨基酸代谢途径。随机森林机器学习方法提取了一些对预测种植年龄或成熟阶段有重要意义的代谢物，错误率低。本研究表明，使用基于 GC-MS 的代谢组学方法和随机森林分析可以成功区分人参果的不同种植年龄或成熟阶段。
