Automated pathway and reaction prediction facilitates in silico identification of unknown metabolites in human cohort studies

Research output: Contribution to journalArticle


Identification of metabolites in non-targeted metabolomics continues to be a bottleneck in metabolomics studies in large human cohorts. Unidentified metabolites frequently emerge in the results of association studies linking metabolite levels to, for example, clinical phenotypes. For further analyses these unknown metabolites must be identified. Current approaches utilize chemical information, such as spectral details and fragmentation characteristics to determine components of unknown metabolites. Here, we propose a systems biology model exploiting the internal correlation structure of metabolite levels in combination with existing biochemical and genetic information to characterize properties of unknown molecules.Levels of 758 metabolites (439 known, 319 unknown) in human blood samples of 2279 subjects were measured using a non-targeted metabolomics platform (LC-MS and GC-MS). We reconstructed the structure of biochemical pathways that are imprinted in these metabolomics data by building an empirical network model based on 1040 significant partial correlations between metabolites. We further added associations of these metabolites to 134 genes from genome-wide association studies as well as reactions and functional relations to genes from the public database Recon 2 to the network model. From the local neighborhood in the network, we were able to predict the pathway annotation of 180 unknown metabolites. Furthermore, we classified 100 pairs of known and unknown and 45 pairs of unknown metabolites to 21 types of reactions based on their mass differences. As a proof of concept, we then looked further into the special case of predicted dehydrogenation reactions leading us to the selection of 39 candidate molecules for 5 unknown metabolites. Finally, we could verify 2 of those candidates by applying LC-MS analyses of commercially available candidate substances. The formerly unknown metabolites X-13891 and X-13069 were shown to be 2-dodecendioic acid and 9-tetradecenoic acid, respectively.Our data-driven approach based on measured metabolite levels and genetic associations as well as information from public resources can be used alone or together with methods utilizing spectral patterns as a complementary, automated and powerful method to characterize unknown metabolites.


  • Jan D. Quell
  • Werner Römisch-Margl
  • Marco Colombo
  • Jan Krumsiek
  • Anne M. Evans
  • Robert Mohney
  • Veikko Salomaa
  • Ulf De Faire
  • Leif C. Groop
  • Felix Agakov
  • Helen C. Looker
  • Paul M. McKeigue
  • Helen M. Colhoun
  • Gabi Kastenmüller
External organisations
  • Helmholtz Zentrum München
  • University of Edinburgh
  • Metabolon Inc.
  • Finnish National Institute for Health and Welfare
  • Karolinska Institutet
  • Technical University of Munich
  • Pharmatics Limited
  • University of Dundee
Research areas and keywords

Subject classification (UKÄ) – MANDATORY

  • Cell and Molecular Biology
  • Medical Laboratory and Measurements Technologies


  • Biochemical pathway prediction, Metabolic network reconstruction, Metabolite identification, Non-targeted metabolomics, Reaction prediction
Original languageEnglish
Pages (from-to)58-67
JournalJournal of Chromatography B: Analytical Technologies in the Biomedical and Life Sciences
Early online date2017 Apr 4
Publication statusPublished - 2017
Publication categoryResearch