Development of a practical metabolite identification technique for non-targeted metabolomics

Metabolite identification is one of the major challenges of non-targeted metabolomics involving liquid chromatography coupled with mass spectrometry (LC-MS). Compound databases contain enormous numbers of records, which makes compound identification difficult in practice because each search will return a large number of candidates. We therefore developed a practical compound identification system using LC-MS with high mass accuracy and MS(n) capability, combined with a compound database. A large number of candidates were evaluated by score calculation based on a combination of formulae and spectral assignments. Here, we demonstrate this method using green tea extract as a model sample. We applied our approach to predict the structures of compounds of interest, and the correct identification of several candidates was confirmed by comparisons to analysis of chemical standards.