Identification of nontarget compounds in complex mixtures is of significant importance in various scientific fields. On the basis of the universal property that the compounds in complex mixtures can be classified to various analogous families, this study presents a general strategy for the rapid identification of nontarget compounds from complex matrixes using herbal medicine as an example. The proposed strategy consists of three sequential steps. First, a blank control sample is prepared for the purpose of removing interferences in the complex matrixes via automatic chromatographic and mass spectrometric data comparisons. Second, the diagnostic ions guided bridging network strategy is developed for the rapid classification of analogous compounds and structural characterizations. Finally, a quantitative structure retention relationship (QSRR) is built to validate the identifications and to differentiate isomers. Using this strategy, we have successfully identified a total of 45 organic acids from Mai-Luo-Ning and Flos Lonicerae injection, and 46 ginsenosides from Shen-Mai injection samples. The QSRR approach enabled a successful differentiation of most isomers. The proposed strategy will be expected to be applicable to the identification of nontarget compounds in complex mixtures.
Keywords: Analogous compounds; BNN; Back Propagation Neural Network; GFA; Genetic Function Approximation; Herbal medicines; LC/MS-IT-TOF; MLR; MS; Multiple Linear Regression; Nontarget analysis and Quantitative structure-retention relationship (QSRR); Organic acids; QSRR; SPE; mass spectrometry; quantitative structure retention relationship; solid phase extraction.
Copyright © 2013 Elsevier B.V. All rights reserved.