Metabolomics includes a variety of applications, many of which ultimately end up with structural assignments of individual metabolic components. To meet the structural challenges set by metabolomics, we have designed a concept for the mapping and spectral characterization of the human metabolome (http://www.highchem.com/metabolome). This concept poses an enormous analytical challenge regarding both the experimental and data processing elements. An integrated approach is being developed, which will address acquisition, processing and, above all, the interpretation of the vast amount of mass spectrometric data needed to accomplish the ultimate goal of creating searchable libraries of detectable human metabolome.

The application of high-resolution, multi-stage tandem mass spectrometry is essential for a structural work of this extent. Data from complex biological samples can be generated in terabyte volumes at a high-throughput speed using a combination of hyphenated techniques and HR mass spectrometry. However, contemporary interpretation techniques require human interaction, supplementary structural information and/or the reference spectra of structurally related compounds to unambiguously determine the structural arrangement of a metabolite. A novel expert system based concept, together with various empirical data collections, is being developed to address these challenges. This concept combines several interconnected platforms which perform systematic spectra interpretation of experimental data. To perform the initial training of the expert system and spectrally characterize metabolites that are commercially available as catalogue chemicals their spectra are being continuously acquired. A complementary approach to this is the identification of known metabolites in real samples based on the exact monoisotopic mass of quasi-molecular, aduct or cluster ions, and consistency checking between predicted and observed fragmentation patterns of known metabolites. For this purpose, we are collating metabolic structural data of human and a number of other eucaryotes from the literature and publicly available libraries, and building an extensive library of fragmentation mechanisms. The spectral data of various eucaryotes promises to considerably enhance the performance of the expert system, as metabolic constituents are generally conserved among the species and thus present the metabolome as a continuum in structural space, in which structures share similar scaffolds. Spectra interpretation of unknown components is accomplished using a trained expert system based on spectral analogies, automated fragmentation prediction and several other advanced algorithms that will be discussed.

An important part of this project is the management of heterogeneous spectral, structural, mechanistic and biochemical data. We employ proven relational database technology to create spectral libraries of metabolic components based on our novel spectral tree representation, which best reflects the hierarchical ion dependencies and addresses the reproducibility of ion ratios in CID experiments. The identified metabolites will be continuously associated with known metabolic pathways using biochemical software developed in-house and based on the collective relational database. The library of fragmentation mechanisms will also share the database platform utilizing the structural protocol common with spectral library and metabolic pathways.

The comprehensive identification of endogenous metabolites and the creation of their spectral signatures promises to have a significant impact on metabolomic research. Instantaneous identification and selective quantitation of metabolic components will facilitate the routine application of mass spectrometry in many fields of metabolomics.