The set of low-molecular weight (usually < 1500 Da) molecules of an organism, organ or tissue is referred to as the metabolome , and the comprehensive qualitative and quantitative analysis of this set of molecules is called metabolomics . Metabolome analyses aim to provide a holistic view of biochemical status at various levels of complexity, from the whole organism, organ or tissue, to the cell, at a given time. Metabolomics is increasingly widely used by plant biologists [3–6] studying the effects of genotype and biotic or abiotic environments [7–9] or the biochemical modifications associated with developmental changes [10, 11]. It is also widely used by food scientists, for descriptions of changes in the organoleptic properties and nutritional quality of food  and evaluations of food authenticity . It is also used in substantial equivalence studies for genetically modified organisms . Metabolomics has also increasingly entered into routine use in plant functional genomics, in which correlations between such biochemical information and genetic and molecular data are improving our insight into the functions of unknown genes [15–17]. Finally, it is emerging as a tool for the screening of genetic resources and plant breeding [18, 19].
The chemical diversity and complexity of the plant metabolome constitutes a real challenge, even for a given species, because the diversity of metabolites and their concentration ranges remains huge. It is therefore impossible to profile all metabolite families (the list of these families includes amino acids, organic acids, carbohydrates, lipids and diverse secondary metabolites, such as phenylpropanoids, isoprenoids, terpenoids and alkaloids) simultaneously through a single extraction and with only one analytical technique. Most metabolomics projects therefore use several analytical strategies in parallel [17, 20]. Several techniques of choice have emerged, including gas chromatography or liquid chromatography coupled with mass spectrometry (GC-MS or LC-MS) and proton nuclear magnetic resonance spectrometry (1H-NMR) [21, 22].
1H-NMR and GC-MS have been applied to polar extracts for the study of primary metabolites. 1H-NMR technology has been widely used as a high-throughput technique for non targeted fingerprinting with little or no sample preparation [23, 24]. It has also been applied to targeted profiling and the absolute quantification of major metabolites , despite its relatively low sensitivity, taking advantage of its large dynamic range . GC-MS is much more sensitive than 1H-NMR and is ideal for the detection of volatile metabolites, but high-boiling point metabolites require two-step derivatization .
The relative quantification of a hundred hydrophilic metabolites can be achieved, but comparisons of sets of GC-MS metabolomics profiles obtained in different laboratories remain difficult. For the study of secondary metabolites, LC-MS analysis is generally the method of choice. Extracts are injected directly, without derivatization. LC-MS is generally used for metabolomic profiling  with relative quantification. The use of shared databases is hindered by cross-compatibility problems between spectra acquired with different LC-MS instruments , even with two instruments of the same model from the same manufacturer. High-resolution MS techniques, such as FT-ICR-MS, are also used without LC separation and are very promising for use in plant metabolomics . However, a complementary technique, such as NMR, is often required for further characterization of specific metabolome changes in terms of structure . A major advantage of 1H-NMR is that the profiles obtained are often comparable, even between different instruments or different field magnitudes [31, 32], provided that some parameters, such as extract pH, are fixed at a constant value.
Metabolomics facilities, including those using 1H-NMR, generate large amounts of raw, processed and analyzed data, which must be well managed if they are to generate useful knowledge. Various web-based software platforms are available for managing and making use of metabolomics data. These software platforms include metabolite spectral databases, such as the Golm Metabolome Database (GMD) and the Human Metabolome DataBase (HMDB). The GMD  provides public access to GC-MS data and peak lists for plant metabolites. The HMDB [33, 34] is an example of an organism-specific database providing detailed information, including quantification and information about the spatial distribution of small metabolites in the human body. These metabolite-oriented platforms also provide simple query forms for searches by mass or compound names. Standard compound libraries, such as the Biological Magnetic Resonance data Bank (BMRB)  are also useful for metabolite identification by NMR. Databases of this type may be seen as knowledgebases rather than integrated tools for data management, analysis and metabolite identification. MeltDB  and SetupX , two web-based software platforms for the systematic storage, analysis and annotation of datasets from mass spectrometry (MS)-based metabolomics experiments, have recently been implemented. However, these platforms cannot handle NMR data. Another platform, PRIMe , provides standardized measurements of metabolites by multidimensional NMR spectroscopy, GC-MS, LC-MS and capillary electrophoresis coupled with MS (CE-MS). It also provides unique tools for metabolomics, transcriptomics and the integrated analysis of a range of other "-omics" data. The standardized spectrum search in PRIMe is a very useful tool, but it does not provide information about the biological context of compounds, unlike the KNApSAcK database linking metabolites identified by MS to species http://www.metabolome.jp/software/knapsack-database or Phenolexplorer , a bibliographic database http://www.phenol-explorer.eu dedicated to the polyphenol content of food. MetaboAnalyst  is an online tool for processing high-throughput metabolomic data from NMR and GC/LC-MS spectra. For NMR, it allows statistical analysis of compound concentration data obtained by quantitative metabolic profiling or of 1H NMR spectral signatures (after data reduction with bucketing) for urine samples for example. MetaboAnalyst does not handle NMR spectra but only processed data (peak list or buckets list) in tabular csv files. Each of these applications is useful, but none constitutes a complete tool for managing, analyzing and sharing plant NMR metabolomics data.
Given the types of metabolomics resources available (listed in ), and the key aspects of both the analysis and understanding of metabolomics data (identified as Visualization in ), there is currently a need for i) a spectral database combined with ii) a knowledgebase for plants, iii) an easy-to-use metabolomic spectral visualization tool and iv) a metabolomic data analysis tool. Taking these requirements into account, we have developed a plant metabolomics platform (with public or private access) for the storage, management, visualization, analysis, annotation and query of NMR fingerprints or quantitative profiles and quantified metabolite. This platform has been named MeRy-B, for Metabolomics Repository Bordeaux. MeRy-B facilitates profile discrimination through the visualization of spectral data by either modular spectrum overlay (i.e. driven by the choice of criteria or factors from the experimental design) or multivariate statistical analysis. It can also construct a knowledgebase of plant metabolites determined by NMR, including metabolite concentration data when available, with minimal information about experimental conditions in the context of scientific publications, and can be used for the re-analysis of combined experiments. Furthermore, MeRy-B provides tools for the identification of metabolites by comparisons of spectra for plant extracts with spectra available in the MeRy-B knowledgebase.