An improved detection and identification strategy for untargeted metabolomics based on UPLC-MS

Untargeted metabolomics provides a comprehensive investigation of metabolites and enables the discovery of biomarkers. Improvements in sample preparation, chromatographic separation and raw data processing procedure greatly enhance the metabolome coverage. In addition, database-dependent software identification is also essential, upon which enhances the identification confidence and benefits downstream biological analysis. Herein, we developed an improved detection and identification strategy for untargeted metabolomics based on UPLC-MS. In this work, sample preparation was optimized by considering chemical properties of different metabolites. Chromatographic separation was done by two different columns and MS detection was performed under positive and negative ion modes regarding to the different polarities of metabolites. According to the characteristics of the collected data, an improved identification and evaluation strategy was developed involving fragment simulation and MS/MS library search based on two commonly used databases, HMDB and METLIN. Such combination integrated information from different databases and was aimed to enhance identification confidence by considering the rationality of fragmentation, biological sources and functions comprehensively. In addition, decision tree analysis and lab-developed database were also introduced to assist the data processing and enhance the identification confidence. Finally, the feasibility of the developed strategy was validated by liver samples of obesity mice and controls. 238 metabolites were accurately detected, which was beneficial for the subsequent biomarker discovery and downstream pathway analysis. Therefore, the developed strategy remarkably facilitated the identification accuracy and the confirmation of metabolites in untargeted metabolomics. (C) 2020 Elsevier B.V. All rights reserved.