Integrative analysis of Cichorium intybus L. Extracts: from NADES-Based extraction and UHPLC-Orbitrap®-HRMS to bioinformatics and machine learning

English (OpenAlex)

Abstract This study aimed to comprehensively evaluate the phytochemical composition, antioxidant, and antimicrobial activities plans aerial parts of Cichorium intybus L. extracts obtained using conventional solvents (dichloromethane, n-hexane, methanol, ethanol) and a natural deep eutectic solvent (NADES). Antioxidant capacity was assessed through DPPH, ABTS radical scavenging, and Fe 3+ /Cu 2+ reducing assays, while antimicrobial potential was evaluated using the microdilution method against Staphylococcus aureus , Bacillus subtilis , and Klebsiella pneumoniae. Chemical profiling of methanol and NADES extracts was performed via UHPLC-Orbitrap ® -HRMS/MS, revealing that NADES extraction enriched glucuronidated flavonoids and hydroxycinnamic acids, including luteolin-7-O-glucuronide, quinic acid, and chlorogenic acid. Chemometric approaches (principal component analysis and heatmap clustering) confirmed distinct metabolite enrichment patterns between extracts. To further elucidate biological relevance, bioinformatic analyses (SwissTargetPrediction, Kyoto Encyclopedia of Genes and Genomes (KEGG), DisGeNET, and GO enrichment) suggested potential molecular targets and pathways linked to antioxidant and antimicrobial activity. Moreover, a machine learning model (Random Forest with leave-one-out cross-validation) combined with Shapley additive explanations (SHAP) identified quinic acid and flavonoid glucuronides as the most predictive determinants of antioxidant capacity, demonstrating > 92% variance explanation with minimal error (R 2 = 0.9225, RMSE = 0.0219). Collectively, these integrative findings highlight NADES as a sustainable extraction medium and underscore the utility of combining chemometrics, bioinformatics, and machine learning for a systems-level understanding of plant-derived bioactive compounds.