Liu Wenbin, Zhang Bingyang, Xin Zhongquan, Ren Dabing, Yi Lunzhao
Yunnan Food Safety Research Institute, Kunming University of Science and Technology, Kunming 650500, China.
School of Science, Kunming University of Science and Technology, Kunming 650500, China.
Int J Mol Sci. 2017 Jul 3;18(7):1342. doi: 10.3390/ijms18071342.
This present study aims to identify the key bioactive components in (ATR), a traditional Chinese medicine (TCM) with various bioactivities. Partial least squares regression (PLSR) was employed to describe the relationship between the radical scavenging activity and the volatile components. The PLSR model was improved by outlier elimination and variable selection and was evaluated by 10-fold cross-validation and external validation in this study. Based on the PLSR model, eleven chemical components were identified as the key bioactive components by variable importance in projection. The final PLS regression model with these components has good predictive ability. The ² was 0.8284, and the root mean square error for prediction was 2.9641. The results indicated that the eleven components could be a pattern to predict the radical scavenging activity of ATR. In addition, we did not find any specific relationship between the radical scavenging ability and the habitat of the ATRs. This study proposed an efficient strategy to predict bioactive components using the combination of quantitative chromatography fingerprints and PLS regression, and has potential perspective for screening bioactive components in complex analytical systems, such as TCM.
本研究旨在鉴定具有多种生物活性的中药蒺藜（ATR）中的关键生物活性成分。采用偏最小二乘回归（PLSR）来描述自由基清除活性与挥发性成分之间的关系。本研究通过异常值消除和变量选择对PLSR模型进行了改进，并通过10倍交叉验证和外部验证对其进行了评估。基于PLSR模型，通过投影变量重要性鉴定出11种化学成分作为关键生物活性成分。包含这些成分的最终PLS回归模型具有良好的预测能力。R²为0.8284，预测均方根误差为2.9641。结果表明，这11种成分可作为预测ATR自由基清除活性的一种模式。此外，我们未发现ATR的自由基清除能力与其产地之间存在任何特定关系。本研究提出了一种结合定量色谱指纹图谱和PLS回归来预测生物活性成分的有效策略，在筛选复杂分析体系（如中药）中的生物活性成分方面具有潜在前景。
