Symposium Sessions (Day1, Day2, Day3)
Basic Sessions (Day1, Day2, Day3)
Young Researchers' Sessions (Day1, Day2, Day3)
Young Researchers' Sessions
- Day 2, May 16（Tue.） 10:00-10:15 Room C (Room 1009)
-
2C-O-1000 PDF
Evaluation of Unified-HILIC/AEX Retention Time Prediction and Structural Annotation
Recently, we developed a unified hydrophilic-interaction/anion-exchange liquid chromatography tandem mass spectrometry (unified-HILIC/AEX/MS) technique that can comprehensively and simultaneously separate and detect a wide range of hydrophilic metabolites. However, retention time (RT) prediction of metabolite peaks detected by this method has not been developed due to its complex separation mechanism, which involves the continuous and stepwise action of the HILIC and AEX modes. In this study, an RT prediction model was developed focusing on three points: (1) acquisition and use of a high-quality training data set, (2) collection of 12,420 molecular descriptors (explanatory variables), and (3) selection of molecular descriptors and detailed model construction. The results showed that the average absolute error value of the model relative to the test data was ±0.796 min, 86% of the ΔRT was within ±1.5 min, and the R2 between measured and predicted RTs was 0.92. Using this method, we successfully annotated 216 metabolic features in nontargeted metabolomics data of human plasma (SRM1950) extract through structural annotation via in silico MS/MS prediction and RT prediction. In particular, the number of candidate structures assigned by in silico MS/MS prediction was reduced to approximately 50% by the addition of RT prediction information. These results demonstrate the usefulness of the unified-HILIC/AEX RT prediction model for comprehensive structural annotation of unknown hydrophilic metabolites.