Discovering Regulated Metabolite Families in Untargeted Metabolomics Studies

The
identification of metabolites by mass spectrometry constitutes
a major bottleneck which considerably limits the throughput of metabolomics
studies in biomedical or plant research. Here, we present a novel
approach to analyze metabolomics data from untargeted, data-independent
LC-MS/MS measurements. By integrated analysis of MS1 abundances
and MS/MS spectra, the identification of regulated metabolite families
is achieved. This approach offers a global view on metabolic regulation
in comparative metabolomics. We implemented our approach in the web
application “MetFamily”, which is freely available at http://msbi.ipb-halle.de/MetFamily/. MetFamily provides a dynamic
link between the patterns based on MS1-signal intensity
and the corresponding structural similarity at the MS/MS level. Structurally
related metabolites are annotated as metabolite families based on
a hierarchical cluster analysis of measured MS/MS spectra. Joint examination
with principal component analysis of MS1 patterns, where
this annotation is preserved in the loadings, facilitates the interpretation
of comparative metabolomics data at the level of metabolite families.
As a proof of concept, we identified two trichome-specific metabolite
families from wild-type tomato Solanum habrochaites LA1777 in a fully unsupervised manner and validated our findings
based on earlier publications and with NMR.