Integrating Metabolomics and Machine Learning to Analyze Chemical Markers and Ecological Regulatory Mechanisms of Geographical Differentiation in Thesium chinense Turcz

• Published in: Metabolites Vol. 15; no. 7; p. 423
• Main Authors: Wang, Cong, Che, Ke, Zhang, Guanglei, Yu, Hao, Wang, Junsong
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 20.06.2025; MDPI
• Subjects: 1-Phosphatidylinositol 3-kinase; AKT protein; Algorithms; Analysis; antioxidant activity; Antioxidants; Authenticity; Biological activity; Biomarkers; Correlation analysis; Data mining; environmental factors; Factor analysis; Flavonoids; geo-authentic herbs; Geographical distribution; Geographical variations; Herbal medicine; Herbs; Hydroxyl radicals; Isoflavones; Learning algorithms; Low temperature; Machine learning; Mass spectrometry; Metabolic pathways; Metabolites; Metabolomics; Nitrogen; Plant metabolites; Precipitation; Quality control; Scientific imaging; Secondary metabolites; Software; Thesium; Thesium chinense Turcz; China; UHPLC-Q-TOF/MS; geo-authentic herbs; Thesium chinense Turcz; environmental factors; metabolomics; machine learning; antioxidant activity
• ISSN: 2218-1989; 2218-1989
• DOI: 10.3390/metabo15070423

Background: The relationship between medicinal efficacy and the geographical environment in Thesium chinense Turcz. (T. chinense Turcz.), a traditional Chinese herb, remains systematically unexplored. This study integrates metabolomics, machine learning, and ecological factor analysis to elucidate the geographical variation patterns and regulatory mechanisms of secondary metabolites in T. chinense Turcz. from Anhui, Henan, and Shanxi Provinces. Methods: Metabolomic profiling was conducted on T. chinense Turcz. samples collected from three geographical origins across Anhui, Henan, and Shanxi Provinces. Machine learning algorithms (Random Forest, LASSO regression) identified region-specific biomarkers through intersection analysis. Metabolic pathway enrichment employed MetaboAnalyst 5.0 with target prediction. Antioxidant activity (DPPH/hydroxyl radical scavenging) was quantified spectrophotometrically. Environmental correlation analysis incorporated 19 WorldClim variables using redundancy analysis, Mantel tests, and Pearson correlations. Results: We identified 43 geographical marker compounds (primarily flavonoids and alkaloids). Random forest and LASSO regression algorithms determined core markers for each production area: Anhui (4 markers), Henan (6 markers), and Shanxi (3 markers). Metabolic pathway enrichment analysis revealed these markers exert pharmacological effects through neuroactive ligand–receptor interaction and PI3K-Akt signaling pathways. Redundancy analysis demonstrated Anhui samples exhibited significantly higher antioxidant activity (DPPH and hydroxyl radical scavenging rates) than other regions, strongly correlating with stable low-temperature environments (annual mean temperature) and precipitation patterns. Conclusions: This study established the first geo-specific molecular marker system for T. chinense Turcz., demonstrating that the geographical environment critically influences metabolic profiles and bioactivity. Findings provide a scientific basis for quality control standards of geo-authentic herbs and offer insights into plant–environment interactions for sustainable cultivation practices.