Advanced Techniques and AI for Identifying Quality Markers in St. John's Wort

Jenn Hoskins
16th June, 2024

Advanced Techniques and AI for Identifying Quality Markers in St. John's Wort

Image Source: Natural Science News, 2024

Key Findings

  • Researchers at Tianjin University identified 122 compounds in St. John's wort using advanced chemical analysis
  • They found 46 key metabolites that differ between plant parts like flowers, leaves, and branches
  • Five specific compounds were selected as quality markers to ensure consistent product quality using machine learning techniques
Hypericum perforatum L. (HPL), commonly known as St. John's wort, is a well-researched medicinal plant renowned for its potential therapeutic benefits, particularly in treating mild to moderate depression. The need for consistent quality control in HPL products is critical due to the variability in chemical composition influenced by factors such as geographical origin, plant part harvested, and processing techniques[2]. In a recent study conducted by Tianjin University of Traditional Chinese Medicine[1], researchers employed non-targeted metabolomics combined with machine learning to identify quality indicators for the holistic quality control of HPL. The study began by collecting high-resolution mass spectrometry (MS) data from various HPL samples and visualizing the chemical compounds through an MS molecular network. This approach led to the identification of 122 compounds. To compare the differences in metabolite expression between different parts of the plant (flower, leaf, and branches), an orthogonal partial least squares-discriminant analysis (OPLS-DA) model was established. This model highlighted 46 differential metabolites. Further analysis focused on the pharmacological activities of these differential metabolites using a protein-protein interaction (PPI) network. This analysis retrieved 25 compounds associated with 473 gene targets. Among these, 13 highly active compounds were identified as potential quality markers, and ultimately, five compounds were selected as quality control markers for HPL. These quality control markers were validated using three different machine learning classifiers: support vector machine (SVM), random forest (RF), and K-nearest neighbor (KNN). The RF model demonstrated optimal performance when the feature count was set to 122 and 46, although its performance degraded as the number of variables decreased. Both the KNN and SVM models also showed a decrease in performance but still met the intended requirements. This study ties together previous findings on the chemical variability and quality control of HPL. For instance, earlier research highlighted that the chemical composition of HPL can vary significantly based on geographical origin, plant part harvested, and processing techniques[2]. This new study reinforces these findings by demonstrating that different parts of the HPL plant exhibit distinct metabolite profiles, which can be effectively identified and analyzed using advanced techniques like non-targeted metabolomics and machine learning. Moreover, the study's findings align with previous research on the accumulation of secondary metabolites in Hypericum species. For example, a prior study found that metabolites such as rutin, hyperoside, quercetin, and hypericin are more accumulated in the leaves than in the stems, while epicatechin showed the opposite trend[3]. The new study's detailed metabolomic analysis provides a more comprehensive understanding of these differences, further supporting the importance of selecting appropriate plant parts for quality control. The innovative approach of combining non-targeted metabolomics with machine learning offers a robust strategy for the quality control of HPL. By identifying specific quality control markers, this method ensures the consistent composition and efficacy of HPL products, addressing the variability issues highlighted in earlier studies[2]. This strategy can also serve as a reference for the quality control of other herbal medicines, potentially improving the overall reliability and safety of herbal products in the market. In conclusion, the study conducted by Tianjin University of Traditional Chinese Medicine provides a significant advancement in the quality control of HPL through the integration of non-targeted metabolomics and machine learning. This approach not only validates previous findings on the chemical variability of HPL but also offers a practical solution for ensuring the consistent quality of HPL products, ultimately benefiting consumers and healthcare providers alike.

MedicineBiotechPlant Science

References

Main Study

1) UPLC-Q-TOF-MS/MS combined with machine learning methods for screening quality indicators of Hypericum perforatum L.

Published 13th June, 2024

https://doi.org/10.1016/j.jpba.2024.116313

Related Studies

2) St. John's Wort (Hypericum perforatum) Products - How Variable Is the Primary Material?

https://doi.org/10.3389/fpls.2018.01973

3) Methodological aspects of biologically active compounds quantification in the genus Hypericum.

https://doi.org/10.1016/j.jpba.2018.03.048


Related Articles

An unhandled error has occurred. Reload 🗙