Metabolomics covers a wide range of applications in life sciences, biomedicine, and phytology. Data acquisition (to achieve high coverage and efficiency) and analysis (to pursue good classification) are two key segments involved in metabolomics workflows. Various chemometric approaches utilizing either pattern recognition or machine learning have been employed to separate different groups. However, insufficient feature extraction, inappropriate feature selection, overfitting, or underfitting lead to an insufficient capacity to discriminate plants that are often easily confused. Using two ginseng varieties, namely Panax japonicus (PJ) and Panax japonicus var. major (PJvm), containing the similar ginsenosides, we integrated pseudo-targeted metabolomics and deep neural network (DNN) modeling to achieve accurate species differentiation. A pseudo-targeted metabolomics approach was optimized through data acquisition mode, ion pairs generation, comparison between multiple reaction monitoring (MRM) and scheduled MRM (sMRM), and chromatographic elution gradient. In total, 1980 ion pairs were monitored within 23 min, allowing for the most comprehensive ginseng metabolome analysis. The established DNN model demonstrated excellent classification performance (in terms of accuracy, precision, recall, F1 score, area under the curve, and receiver operating characteristic (ROC)) using the entire metabolome data and feature-selection dataset, exhibiting superior advantages over random forest (RF), support vector machine (SVM), extreme gradient boosting (XGBoost), and multilayer perceptron (MLP). Moreover, DNNs were advantageous for automated feature learning, nonlinear modeling, adaptability, and generalization. This study confirmed practicality of the established strategy for efficient metabolomics data analysis and reliable classification performance even when using small-volume samples. This established approach holds promise for plant metabolomics and is not limited to ginseng.

Identifying the compound formulae-related xenobiotics in bio-samples is full of challenges.Conventional strategies always exhibit the insufficiencies in overall coverage,analytical efficiency,and degree of automation,and the results highly rely on the personal knowledge and experience.The goal of this work was to establish a software-aided approach,by integrating ultra-high performance liquid chromatography/ion-mobility quadrupole time-of-flight mass spectrometry(UHPLC/IM-QTOF-MS)and in-house high-definition MS2 library,to enhance the identification of prototypes and metabolites of the compound formulae in vivo,taking Sishen formula(SSF)as a template.Seven different MS2 acquisition methods were compared,which demonstrated the potency of a hybrid scan approach(namely high-definition data-independent/data-dependent acquisition(HDDIDDA))in the identification precision,MS1 coverage,and MS2 spectra quality.The HDDIDDA data for 55 reference compounds,four component drugs,and SSF,together with the rat bio-samples(e.g.,plasma,urine,feces,liver,and kidney),were acquired.Based on the UNIFI? platform(Waters),the efficient data processing workflows were estab-lished by combining mass defect filtering(MDF)-induced classification,diagnostic product ions(DPIs),and neutral loss filtering(NLF)-dominated structural confirmation.The high-definition MS2 spectral li-braries,dubbed in vitro-SSF and in vivo-SSF,were elaborated,enabling the efficient and automatic identification of SSF-associated xenobiotics in diverse rat bio-samples.Consequently,118 prototypes and 206 metabolites of SSF were identified,with the identification rate reaching 80.51％and 79.61％,respectively.The metabolic pathways mainly involved the oxidation,reduction,hydrolysis,sulfation,methylation,demethylation,acetylation,glucuronidation,and the combined reactions.Conclusively,the proposed strategy can drive the identification of compound formulae-related xenobiotics in vivo in an intelligent manner.