Applied Bioinformatics Group

A   A   A
Home > Publications > Automated Label-Free Quantification of Metabolites from LC-MS Data

Skip to content. | Skip to navigation

Document Actions

Erhan Kenar, Holger Franken, Sara Forcisi, Kilian Wörmann, Hans-Ulrich Häring, Rainer Lehmann, Philippe Schmitt-Kopplin, Andreas Zell, and Oliver Kohlbacher (2014)

Automated Label-Free Quantification of Metabolites from LC-MS Data

Mol. Cell. Prot., 13(1):348-59.

Liquid chromatography coupled to mass spectrometry (LC-MS) has become a standard technology in metabolomics. In particular, label-free quantification based on LC-MS is easily amenable to large-scale studies and is thus easily amenable to clinical metabolomics. Large-scale studies, however, require automated processing of the large and complex LC-MS datasets. We present a novel algorithm for the detection of mass traces and their aggregation into features (i.e., all signals caused by the same analyte species) that is computationally efficient, sensitive, and leads to reproducible quantification results. The algorithm is based on a sensitive detection of mass traces, which are then assembled into features based on mass-to-charge spacing, co-elution information, and a support vector machine (SVM)-based classifier able to identify potential metabolite isotope patterns. The algorithm is not limited to metabolites but is applicable to a wide range of small molecules (e.g., lipidomics, peptidomics) as well as to other separation technologies (e.g., CE-MS). We assess the algorithm’s robustness to varying noise levels on synthetic data and then validate the approach on experimental data investigating human plasma samples. We obtain excellent results in a fully automated data processing pipeline both with respect to accuracy and reproducibility. Compared to state-of-the art algorithms, we can demonstrate increased precision and recall of the method. The algorithm is available as part of the open-source software package OpenMS ( and runs on all major operating systems.