Metabolome searcher: a high throughput tool for metabolite identification and metabolic pathway mapping directly from mass spectrometry and using genome restriction
© Dhanasekaran et al.; licensee BioMed Central. 2015
Received: 4 August 2014
Accepted: 13 January 2015
Published: 25 February 2015
Mass spectrometric analysis of microbial metabolism provides a long list of possible compounds. Restricting the identification of the possible compounds to those produced by the specific organism would benefit the identification process. Currently, identification of mass spectrometry (MS) data is commonly done using empirically derived compound databases. Unfortunately, most databases contain relatively few compounds, leaving long lists of unidentified molecules. Incorporating genome-encoded metabolism enables MS output identification that may not be included in databases. Using an organism’s genome as a database restricts metabolite identification to only those compounds that the organism can produce.
To address the challenge of metabolomic analysis from MS data, a web-based application to directly search genome-constructed metabolic databases was developed. The user query returns a genome-restricted list of possible compound identifications along with the putative metabolic pathways based on the name, formula, SMILES structure, and the compound mass as defined by the user. Multiple queries can be done simultaneously by submitting a text file created by the user or obtained from the MS analysis software. The user can also provide parameters specific to the experiment’s MS analysis conditions, such as mass deviation, adducts, and detection mode during the query so as to provide additional levels of evidence to produce the tentative identification. The query results are provided as an HTML page and downloadable text file of possible compounds that are restricted to a specific genome. Hyperlinks provided in the HTML file connect the user to the curated metabolic databases housed in ProCyc, a Pathway Tools platform, as well as the KEGG Pathway database for visualization and metabolic pathway analysis.
Metabolome Searcher, a web-based tool, facilitates putative compound identification of MS output based on genome-restricted metabolic capability. This enables researchers to rapidly extend the possible identifications of large data sets for metabolites that are not in compound databases. Putative compound names with their associated metabolic pathways from metabolomics data sets are returned to the user for additional biological interpretation and visualization. This novel approach enables compound identification by restricting the possible masses to those encoded in the genome.
Bacterial metabolism impacts almost every aspect of our life. Microbial metabolism was exploited by early human civilization to create fermented foods and beverages [1,2]. The oldest known metabolically derived products from microbes include bread, cured meats, cheese, and beer [2-4]. Currently, metabolic engineering for the production of pharmaceuticals and bioactive compounds is giving way to discovery of novel metabolic pathways for production of alternative fuels [5-7]. Burgeoning needs to produce novel antibiotics for disease treatment and health supplements, such as amino-sugars and vitamins, also represent the metabolic end products that are genome encoded of an organism [8-11].
The virulence of bacterial pathogens is closely linked to their metabolism during infection, which is leading to metabolomic disease biomarkers that is pushing the boundaries of robust methods to quickly identify high throughput metabolomic data [12,13]. Cumulatively, the unusual metabolic networks of organisms in ecological niches are renewing interests in metabolites that highlight the lack of high throughput analysis tools for rapid compound identification when the compound is not included in a database. Unfortunately, rapid identification of multiple metabolites simultaneously is also lacking. However, if one considers an organism’s genome to be a database of possible metabolic pathways and metabolite production, it enables customization of MS output analysis based on a specific organism. Approaching the genome as a metabolite database is being done using metabolic reconstruction methods in KEGG and Pathway Tools.
The metabolism of an organism changes during growth, survival, and persistence via complex gene expression changes. In many cases, metabolism begins with the transport of chemically diverse molecules for integration into biologically functional blocks. An organism’s metabolic capability can be envisaged as a highly interconnected network of enzymatic reactions that provide energy, intermediates for macromolecular biosynthesis, cellular signaling, regulation of stress, and control of oxidation/reduction to ensure growth or survival . Highly tuned regulatory mechanisms to modulate the metabolic network via gene expression and enzyme attenuation are needed to quickly adapt to local environmental changes. Evolution of genetic control and gene acquisition are critical to ensure the organism’s survival in the near- and long-term . Adaptation and genetic evolution results in new metabolic nodes in the interconnected network that modifies the intermediate and end product metabolism [14,16]. Of recent interests, metabolic engineering is largely dependent on understanding the metabolic network to regulate production of specific low molecular weight end products that often accumulate.
Metabolite distribution by molecular mass across metabolic encyclopaedias
Molecular weight range (Da)
Number of compounds in Metacyc
Number of compounds in KEGG
Metabolomics aspires to identify all the metabolites produced by an organism [20,21]. However, large data sets, limited identification databases, and limited MS parameters to differentiate small molecules are stumbling blocks for metabolomic analysis, which in turn limits the subsequent bioinformatic analysis and construction of biologically informative models [16,21]. Currently, NMR is of limited use for high throughput small molecule identification due to the lack of sensitivity and limited throughput, but is useful to elucidate the structures of unique metabolites [22,23]. However, NMR is very useful to track the metabolic fate of a small molecule with isotope labels, which provides information for a handful of metabolites once the entire compound list is narrowed to a specific set of metabolic intermediates . Other post-separation detection techniques like photometric, electrochemical, and fluorescent detection are actively used to identify specific metabolites at a substantially reduced analytical scale, but the need to identify the set of compounds produced is overwhelmingly changing the goals of metabolite analysis [24-26]. Conversely, MS analysis, in addition to metabolic tracking estimates the masses of hundreds to thousands of small molecules within minutes and provides information on their relative levels in the sample [27-29], making it very useful for high throughput metabolome analysis. However, it lacks specific information as to the identity of the small molecules, which highlights the need to have curated databases for compound identification .
One approach to overcome the need to identify important molecules uses principal component analysis (PCA) to find changes with a specific treatment. From MS data acquisition this produces a reduced list of small molecules that are tagged as biomarkers [30,31]. Often the diagnostic peak is an unknown compound that is difficult to identify. Subsequently, more complex chemical analysis is used to determine the elemental composition of these biomarkers, which requires additional time, expertise, and often multiple instrumentation capabilities [32,33]. Biomarker identities are subsequently validated by standard compound injection to produce a compound library . While this statistics-driven analytical approach favors method development for MS it ignores the underlying biochemistry and the importance of relatively minor changes of small molecules that can sometimes lead to misinterpretation of the biological impact with new small molecule production. This is especially prevalent for key metabolite classes like hormones, vitamins, and enzyme co-factors where small changes regulate large scale proteomic and metabolic fluctuations . One way to overcome this limitation is to use tools that include all possible putative compounds generated directly from matched compound identities prior to statistical analysis. Subsequently, a significant list of putative compounds can be used for metabolic mapping to facilitate biological identity by linking compound identities to metabolic pathways and routes. Feist et al.  review the reconstruction approach with specific attention to metabolite identification.
Unfortunately, metabolite identification from hundreds to thousands of masses by searching a large compound database is a slow process that is ill defined relative to the specific search criteria that provides confident compounds assignments. GC-MS analysis often identifies compounds by comparison of MS spectra with large, well-established compound libraries (www.nist.gov). Such compound libraries for LC-MS analysis are available for only a small set of masses and are tightly linked to the LC conditions. Large compound databases such as Pubchem (http://pubchem.ncbi.nlm.nih.gov) and Chemspider (http://www.chemspider.com) allow searches of single masses and other query types, but they do not allow queries from large lists of masses or connect putative compounds to metabolic pathways. However, as the query list expands, as it does in metabolome data sets, data analysis using single queries becomes unrealistic for a timely and accurate analysis.
Multiple software suites are available for compound identification of mass spectrometry-based metabolite data that use mass spectral deconvolution and matching to reference databases. Some examples of full-fledged independent platforms are MetSign , MZmine 2 , MAVEN , and XCMS2 , whereas MS Excel templates such as IDEOM , R packages like AStream  and MAIT , and web-applications like METLIN , XCMS Online , and MZedDB  are also available as web services. These tools offer either statistical or structural analyses of small molecule MS data and extract information from metabolic databases to create a list of compounds for their own localized database. For example, MetSign’s compound database is formed from the cumulative compound collection of KEGG, HMDB, and LIPIDMAPS databases, MZmine 2′s collection is from KEGG , HMP, and Pubchem compound, MAVEN uses KEGG, whereas MAIT, IDEOM and ASTREAM use unspecified databases. However, downstream of compound identification, they ignore the underlying biology and do not offer a mechanism to map the data back to the metabolic pathways. Further, they lack the flexibility of implementing user-defined parameters for database searches, as for example, electrospray ionization (ESI) parameters that are predefined in METLIN and MZedDB .
Querying large compound databases that contain millions of non-biological molecules can impede a researcher’s ability to overlay a metabolic context onto metabolomic data . Biologists are producing data at rates that outstrip the ability of analysts to examine the data set to uncover the biological importance. To keep pace with metabolome analysis, high throughput bioinformatic tools that bring compound identity and pathway relevance together to the biologist are crucial. This can be accomplished with: a) automated searches of metabolic databases to retrieve putative compound identification, b) large scale queries be performed seamlessly with MS output, c) provide users the flexibility of using multiple query types, and d) map query results to metabolic pathways, hence allowing data to be analyzed in a biological context.
The availability of over 1,000 annotated microbial genome sequences enables bioinformatic reconstruction (biocyc.org) of an organism’s metabolic capability via the genome, which provides a broad network of metabolism that can be used to predict small molecule production [27,28]. Consequently, recent efforts have focused on uncovering the metabolic networks in many different biological systems [19,37]. Genome reconstructions of the metabolic pathways coupled to analytical methods, such as liquid chromatography (LC), gas chromatography (GC) and capillary electrophoresis with nuclear magnetic resonance spectroscopy (NMR) and mass spectrometry (MS) produces a new method to leverage genomic sequence to provide putative compound identification quickly [27,38].
In this study, a user-friendly web-based application called Metabolome Searcher to retrieve a list of small molecules identifications based on chemical formula, SMILES structure, and the monoisotopic mass was created using an organism’s genome as a putative compound database. While single queries can be directly entered multiple queries with one or more query types can also be done using a text file containing the query list. One or more reference databases can be selected from the list against which the queries are performed. The output connects small molecules in a sample to metabolic databases via embedded links to specific metabolic pathways. The Metabolome Searcher’s output allows researchers using metabolome data from different technologies to group the compound identifications into metabolic information so as to uncover the relevant biological function with multiple chemical criteria.
The ProCyc webserver
We currently house a metabolic database webserver called ProCyc (www.usu.edu/westcent/procyc), which is an implementation of the Pathway Tools webserver (SRI Bioinformatics, Menlo Park, CA) with our own manually and automatically curated metabolic databases of interest. ProCyc houses over 47 metabolic database reconstructions of different classes of bacteria including probiotics, lactic acid bacteria, pathogens, and environmental bacteria that were reconstructed locally. The MetaCyc database and Human metabolism database are part of the basic installation of Pathway Tools software. Some of the reconstructed databases and the tier I/II databases of the basic software were used to exemplify the Metabolome Searcher implementation. This particular platform was chosen for its flexibility to immediately incorporate user-discovered pathways into the right metabolic databases.
Metabolic reference database creation
Organism-specific and general metabolic reference databases available for the Metabolome Searcher
Metabolic reference database
Escherichia coli K12
E. coli K12
Escherichia coli O157:H7
E. coli O157:H7
Lactococcus lactis ssp. lactis IL1403
L. lactis ssp. lactis IL1403
Lactococcus lactis ssp. cremoris SK11
L. lactis ssp. cremoris SK11
Lactobacillus acidophilus NCFM
Lb. acidophilus NCFM
Lactobacillus johnsonii NCC 533
Lb. johnsonii NCC 533
Lactobacillus plantarum WCFS1
Lb. plantarum WCFS1
Listeria monocytogenes EGDe
Mycobacterium bovis AF2122/97
M. bovis AF2122/97
Staphylococcus aureus Mu50
S. aureus ssp. aureus Mu50
Saccharomyces cerevisiae S288C
S. cerevisiae S288C
Salmonella enterica ssp. enterica serovar Typhimurium LT2
Salmonella typhimurium LT2
Compound identification for MS analysis
For compound identification from monoisotopic masses, the user specifies the acceptable deviation from the theoretical masses (ppm or Da, under “Mass deviation”; Figure 1), the ionization mode (positive or negative, under “Electrospray mode”; Figure 1), the maximum number of charges (0-5; under the “Number of proton charge states”; Figure 1), and adducts (mass or formula; optional; under “Adduct or Deduct molecule” and “Maximum number of adducts/deducts”; Figure 1). The deviation value allows the software to obtain matches for queried masses within an acceptable range to narrow or expand the putative identification list. Acceptable mass deviation values may be experimentally determined or obtained from the literature based on a particular instrument and operating conditions .
Typically during MS analysis the molecules are detected by prior ionization with or by removal of protons (positive and negative mode, respectively) . The MS settings are optimized to mainly produce singly charged ions. However, a molecule may still carry multiple charges depending on the MS settings . The user can verify the charge state of compounds contained in the input list to recalibrate the MS settings by selecting different charge states during multiple search sessions.
Positively charged ionic species, such as sodium (Na+) and potassium (K+), or negative species, such as chloride (Cl−) and formate (HCOO−), are also used during ionization due to their abundance in a sample. The addition of ionic species or adducts during ionization shifts the observed monoisotopic mass from that of the intact molecule plus/minus a proton . These adducts can be specified either as individual elements or as partial functional groups in the “Adduct or Deduct molecule” textbox (Figure 1). Similar to adducts, if the user wishes to specify fragments lost during ionization or fragmentation the “Deduct” option can be selected. The user can also provide more than one adduct or deduct in the textbox simultaneously and specify the number of maximum possible adducts or fragments (“Maximum number of adducts/deducts” option).
MRDBs that contain metabolites from different PGDBs or the KEGG database along with calculated monoisotopic masses are used for the queries. MRDBs are included for user selection from the ones listed on the interface (Table 2) wherein the user can select single or multiple MRDBs for searching (Figure 1). If the user intends to query known metabolic pathways in an organism, the organism-specific MRDBs are provided for more specific and narrow options of possible compounds due to the known annotated pathways. However, if the intent is to discover new pathways unknown in a particular system, but identified in other organisms, or if an organism without a pre-constructed MRDB is being studied, the user can select a genotypically related organism’s MRDB or the MetaCyc MRDB for matching. A user-generated PGDB can also be incorporated as an MRDB using the scripts defined above prior to the user defined query. The MRDBs were created in a flat file format to reduce complexity in processing and data handling such that newer MRDBs for other organisms can always be created in a consistent format and readily incorporated as per the user’s need. Pathway Tools was selected as the main metabolic database platform to create MRDBs and link back to PGDBs due to its interactive features and user-level flexibility for metabolic database development and curation of whole genome PGDBs , while queries of an MRDB for the KEGG database  are also supported.
Three different output files are provided as the result of the analysis, one HTML and two text files. The two text files are embedded as links at the top of the HTML page (Figure 3A) that the user can download. One text file (“compounds file”) lists only the matched compounds without any metabolic pathway information, while the other (“pathways file”) repeats each compound’s data by all the pathways that it belongs to as a metabolite.
All scripts were written in Perl (v5.8.6; www.perl.org). The scripts and the metabolic reference databases for Metabolome Searcher are hosted in an Apple XGrid computational cluster (Panther OS 10.3.9) at the Western Dairy Center at Utah State University as well as University of California, Davis. Web pages for data input and output were created using Perl CGI.
MS data validation
Chemical standards preparation
All compounds used were purchased from Sigma-Aldrich (St. Louis, MO). A chemically defined medium described previously by Ganesan et al.  was prepared as a complex mixture for testing Metabolome Searcher’s performance. The major components of this medium are 20 amino acids, sodium chloride, citrate, phosphate, 3-(N-morpholino) propane sulfonic acid (MOPS), vitamin solution (containing 15 different compounds), and glucose. Individual standard solutions of selected amino acids, glucose, citrate, and MOPS were also used for molecule identification.
Separation and analysis of standard compound mixtures were done at the mass spectrometry facility in the CIB. The samples were separated by liquid chromatography (2795 LC system; Waters) prior to introduction by electrospray into the mass spectrometer (QTof Premier; Waters) as described by Mortishire-Smith et al. . Briefly, the separation was done for 10 min using a linear gradient of water:acetonitrile from 0-95% using a Symmetry C18 column (Waters). After introduction into the MS by electrospray, the molecules were detected using both positive and negative electrospray conditions, with calibrated settings recommended by the manufacturer. The QTof instrument was operated in W mode throughout MS analysis. For both positive and negative electrospray analysis, the conditions were: desolvation temperature of 250°C, source temperature of 120°C, cone voltage of 40 V, and collision energy of 4 eV. Data acquisition was performed for a mass range of 50–1,000 Da. After acquisition the data were centroided  using 1 ng/μl leucine-enkephalin infused at 10 μl/min as a reference, with an m/z of 556.2771 in positive mode and m/z of 554.25 in negative mode. In order to subtract background from the LC column and sample matrix, HPLC-grade water (Thermo Fisher Scientific Inc., Waltham, MA) was injected into the MS as a negative control. All samples were analyzed in technical duplicates.
Peak detection, intensity extraction, and normalization were performed using MarkerLynx software (Waters) to obtain monoisotopic masses and molecule retention times. In this study, only the monoisotopic masses of the markers were used for database searches. The Metabolome Searcher does not support any data analysis of the concentrations or relative measures of compound levels obtained from MarkerLynx.
Results and discussion
Metabolomic assessment provides a list of compounds that facilitates the estimation of metabolic flux through both single pathways and networks [41,42]. Metabolome analysis enables determination of abiotic conditions and genetic regulation of metabolic networks. To achieve these purposes a tool that rapidly determines the compound identity, pathways, and metabolic networks was needed [43,44]. The tool accepts queries from common data types and facilitates data integration from independent sources into a unified compound identification and pathway-mapping scheme. To our knowledge such a tool is not available. The Metabolome Searcher addresses these purposes by receiving input from the user, querying the user-selected metabolic reference database(s), and displaying the generated output for further biological interpretation (Figure 3).
Of the Metabolome Searcher’s outputs, the compounds file is useful when the user plans to conduct compound classification, data clustering, principal component analysis, analysis of variance, or graphical visualization. The pathways file allows the users to sort the data by pathways and facilitates interpretation of metabolic flux and pathway connections to determine if a compound is an intermediate or an end product. The main feature of the HTML output is that it lists and links compounds to all metabolic pathways in which the metabolite is involved (Figure 3). These links help the user understand the role of that particular metabolite in the organism’s metabolic network. The user can click on any one of these links that will navigate them to the PGDBs curated and hosted at ProCyc. The user need not repeat queries on the Metabolome Searcher as the HTML file contains the links to the pathways associated with the returned putative compound IDs. To facilitate obtaining the standard chemicals for verification of retention times, CAS IDs of compounds (where available) are also included in a separate column in the output file.
For names, formulae, and SMILES structures, any partial matches will also be detected and listed. For example, a query of the word string “glucose” against the MetaCyc database will identify D-glucose and an additional 52 hits (data not shown) that also include alpha-methyl-glucose, NDP-Glucoses, and all other molecules that contain the substring “glucose” in the name. String matching offers the user the ability to obtain partial matches and allows additional control over the query specificity and flexibility for unknown pathways. In most cases, if the specific MetaCyc compound names are used, the results will be restricted to one hit. Searching of word strings was implemented in order that even if other data sources such as GC-MS and LC-MS/MS were provided after identification using other software suites, or even data from standard GC or HPLC analyses based on extractions and retention times under certain conditions was provided, the data can be mapped to metabolites and pathways.
Summary of hits to selected compounds from a chemically defined growth medium determined from a query of monoisotopic masses using the Metabolome Searcher
Detected by mass match
Number of additional hits*
Number of non-isomeric additional hits*
Uses of metabolome searcher
The Metabolome Searcher provides an automated tool to identify metabolites from MS analyses from metabolic reconstruction of specific genomes. This approach couples long lists of masses to specific genomic-based metabolites for identification and subsequent visualization via metabolic pathways. The tool is flexible so that query types can use many types of data that include names, molecular formulae, or SMILES structures, and monoisotopic masses that are entered singly or in bulk as a text file. The matches to queries are then presented as results along with other input parameters that the user included in the query and the pathways in which the matched metabolites are involved. The versatility of accepted query types and the provision of pathways mapped to queries are unique to the Metabolome Searcher. The Metabolome Searcher’s utility and flexibility facilitates rapid advances from metabolomics to biological comprehension.
Availability of supporting data
Metabolome Searcher can be accessed at http://procyc.westcent.usu.edu/cgi-bin/MetaboSearcher.cgi.
Funding for this project was provided by a grant from USDA CSREES 2006-34526-17001 to BCW. Contribution Number 8121of the Utah Agricultural Experimental Station.
- Kilara A, Shahani KM. Lactic fermentation of dairy foods and their biological significance. J Dairy Sci. 1978;61:1793–800.View ArticleGoogle Scholar
- Sandine WE, Elliker PR. Microbially induced flavours and fermented food flavour in fermented dairy products. J Agr Food Chem. 1970;18:557–62.View ArticleGoogle Scholar
- Dickinson JR, Harrison SJ, Hewlins MJE. An investigation of the metabolism of valine to isobutyl alcohol in Saccharomyces cerevisiae. J Biol Chem. 1998;273(40):25751–6.View ArticlePubMedGoogle Scholar
- Dickinson JR, Lanterman MM, Danner DJ, Pearson BM, Sanz P, Harrison SJ, et al. A 13C nuclear magnetic resonance investigation of the metabolism of leucine to isoamyl alcohol in Saccharomyces cerevisiae. J Biol Chem. 1997;272(43):26871–8.View ArticlePubMedGoogle Scholar
- Li Y, Horsman M, Wu N, Lan CQ, Dubois-Calero N. Biofuels from Microalgae. Biotechnol Prog. 2008;24(4):815–20. doi:10.1021/bp070371k.PubMedGoogle Scholar
- Singh OV, Harvey SP. Integrating biological processes to facilitate the generation of ‘Biofuel’. J Ind Microbiol Biotechnol. 2008;35(5):291–2.View ArticlePubMedGoogle Scholar
- Vertes AA, Inui M, Yukawa H. Technological options for biological fuel ethanol. J Mol Microbiol Biotechnol. 2008;15(1):16–30.View ArticlePubMedGoogle Scholar
- Coates PM. Dietary supplements and health: the research agenda. Novartis Found Symp. 2007;282:202–7. discussion 207-218.View ArticlePubMedGoogle Scholar
- Crowder MW, Spencer J, Vila AJ. Metallo-beta-lactamases: novel weaponry for antibiotic resistance in bacteria. Acc Chem Res. 2006;39(10):721–8.View ArticlePubMedGoogle Scholar
- Roghmann MC, McGrail L. Novel ways of preventing antibiotic-resistant infections: what might the future hold? Am J Infect Control. 2006;34(8):469–75.View ArticlePubMedGoogle Scholar
- Yoneyama H, Katsumata R. Antibiotic resistance in bacteria and its future for novel antibiotic development. Biosci Biotechnol Biochem. 2006;70(5):1060–75.View ArticlePubMedGoogle Scholar
- Deutscher J, Herro R, Bourand A, Mijakovic I, Poncet S. P-Ser-HPr–a link between carbon metabolism and the virulence of some pathogenic bacteria. Biochim Biophys Acta. 2005;1754(1–2):118–25.View ArticlePubMedGoogle Scholar
- Lemercier G, Espiau B, Ruiz FA, Vieira M, Luo S, Baltz T, et al. A pyrophosphatase regulating polyphosphate metabolism in acidocalcisomes is essential for Trypanosoma brucei virulence in mice. J Biol Chem. 2004;279(5):3420–5.View ArticlePubMedGoogle Scholar
- Moxley JF, Jewett MC, Antoniewicz MR, Villas-Boas SG, Alper H, Wheeler RT, et al. Linking high-resolution metabolic flux phenotypes and transcriptional regulation in yeast modulated by the global regulator Gcn4p. Proc Natl Acad Sci U S A. 2009;106(16):6477–82.View ArticlePubMedPubMed CentralGoogle Scholar
- Taylor BL, Zhulin IB. In search of higher energy: metabolism-dependent behaviour in bacteria. Mol Microbiol. 1998;28(4):683–90.View ArticlePubMedGoogle Scholar
- Nedderman AN. Metabolites in safety testing: metabolite identification strategies in discovery and development. Biopharm Drug Dispos. 2009;30(4):153–62.View ArticlePubMedGoogle Scholar
- Krieger CJ, Zhang P, Mueller LA, Wang A, Paley S, Arnaud M, et al. MetaCyc: a multiorganism database of metabolic pathways and enzymes. Nucleic Acids Res. 2004;32(Database issue):D438–42.View ArticlePubMedPubMed CentralGoogle Scholar
- Ganesan B, Dobrowolski P, Weimer B. C.: Identification of the catabolic pathway of leucine-to-2-methylbutyric acid by Lactococcus lactis. Appl Environ Microbiol. 2006;72(6):4264–73.View ArticlePubMedPubMed CentralGoogle Scholar
- Novak L, Loubierre P. The metabolic network of Lactococcus lactis: distribution of 14 C-labelled substrates between catabolic and anabolic pathways. J Bacteriol. 2000;182(4):1136–43.View ArticlePubMedPubMed CentralGoogle Scholar
- Brown M, Dunn WB, Dobson P, Patel Y, Winder CL, Francis-McIntyre S, et al. Mass spectrometry tools and metabolite-specific databases for molecular identification in metabolomics. Analyst. 2009;134(7):1322–32.View ArticlePubMedGoogle Scholar
- Dettmer K, Aronov PA, Hammock BD. Mass spectrometry-based metabolomics. Mass Spectrom Rev. 2007;26(1):51–78.View ArticlePubMedPubMed CentralGoogle Scholar
- Parr AJ, Mellon FA, Colquhoun IJ, Davies HV. Dihydrocaffeoyl polyamines (kukoamine and allies) in potato (Solanum tuberosum) tubers detected during metabolite profiling. J Agric Food Chem. 2005;53(13):5461–6.View ArticlePubMedGoogle Scholar
- Whitfield PD, German AJ, Noble PJ. Metabolomics: an emerging post-genomic tool for nutrition. Br J Nutr. 2004;92(4):549–55.View ArticlePubMedGoogle Scholar
- Feist AM, Herrgard MJ, Thiele I, Reed JL, Palsson BO. Reconstruction of biochemical networks in microorganisms. Nat Rev. 2009;7(2):129–43.Google Scholar
- Wei X, Sun W, Shi X, Koo I, Wang B, Zhang J, et al. MetSign: a computational platform for high-resolution mass spectrometry-based metabolomics. Anal Chem. 2011;83(20):7668–75.View ArticlePubMedPubMed CentralGoogle Scholar
- Pluskal T, Castillo S, Villar-Briones A, Oresic M. MZmine 2: modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profile data. BMC Bioinformatics. 2010;11(1):395.View ArticlePubMedPubMed CentralGoogle Scholar
- Clasquin MF, Melamud E, Rabinowitz JD. LC-MS Data Processing with MAVEN: A Metabolomic Analysis and Visualization Engine. Curr Protoc Bioinformatics. 2012;37:14.11.11– 23.Google Scholar
- Benton HP, Wong DM, Trauger SA, Siuzdak G. XCMS2: processing tandem mass spectrometry data for metabolite identification and structural characterization. Anal Chem. 2008;80(16):6382–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Creek DJ, Jankevics A, Burgess KEV, Breitling R, Barrett MP. IDEOM: an excel interface for analysis of LC–MS-based metabolomics data. Bioinformatics. 2012;28(7):1048–9.View ArticlePubMedGoogle Scholar
- Alonso A, Julià A, Beltran A, Vinaixa M, Díaz M, Ibañez L, et al. AStream: an R package for annotating LC/MS metabolomic data. Bioinformatics. 2011;27(9):1339–40.View ArticlePubMedGoogle Scholar
- Fernández-Albert F, Llorach R, Andrés-Lacueva C, Perera A. An R package to analyse LC/MS metabolomic data: MAIT (Metabolite Automatic Identification Toolkit). Bioinformatics. 2014;30(13):1937–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Smith CA, Maille GO, Want EJ, Qin C, Trauger SA, Brandon TR, et al. METLIN: a metabolite mass spectral database. Ther Drug Monit. 2005;27(6):747–51.View ArticlePubMedGoogle Scholar
- Tautenhahn R, Patti GJ, Rinehart D, Siuzdak G. XCMS Online: a web-based platform to process untargeted metabolomic data. Anal Chem. 2012;84(11):5035–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Draper J, Enot DP, Parker D, Beckmann M, Snowdon S, Lin W, et al. Metabolite signal identification in accurate mass metabolomics data with MZedDB, an interactive m/z annotation tool utilising predicted ionisation behaviour ‘rules’. BMC Bioinformatics. 2009;10:227.View ArticlePubMedPubMed CentralGoogle Scholar
- Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M. The KEGG resource for deciphering the genome. Nucleic Acids Res. 2004;32(Database issue):D277–80.View ArticlePubMedPubMed CentralGoogle Scholar
- Sumner LW, Urbanczyk-Wochniak E, Broeckling CD. Metabolomics data analysis, visualization, and integration. Methods Mol Biol. 2007;406:409–36.PubMedGoogle Scholar
- Calbiani F, Careri M, Elviri L, Mangia A, Zagnoni I. Matrix effects on accurate mass measurements of low-molecular weight compounds using liquid chromatography-electrospray-quadrupole time-of-flight mass spectrometry. J Mass Spectrom. 2006;41(3):289–94.View ArticlePubMedGoogle Scholar
- Kevin Schug HMM. Adduct formation in electrospray ionization. Part 1: common acidic pharmaceuticals. J Sep Sci. 2002;25(12):759–66.View ArticleGoogle Scholar
- Kanehisa M. The KEGG database. Novartis Found Symp. 2002;247:91–101. discussion 101–103, 119–128, 244–152.View ArticlePubMedGoogle Scholar
- Mortishire-Smith RJ, O’Connor D, Castro-Perez JM, Kirby J. Accelerated throughput metabolic route screening in early drug discovery using high-resolution liquid chromatography/quadrupole time-of-flight mass spectrometry and automated data analysis. Rapid Commun Mass Spectrom. 2005;19(18):2659–70.View ArticlePubMedGoogle Scholar
- Ando S, Tanaka Y. Mass spectrometric studies on brain metabolism, using stable isotopes. Mass Spectrom Rev. 2005;24(6):865–86.View ArticlePubMedGoogle Scholar
- Baverel G, Conjard A, Chauvin MF, Vercoutere B, Vittorelli A, Dubourg L, et al. Carbon 13 NMR spectroscopy: a powerful tool for studying renal metabolism. Biochimie. 2003;85(9):863–71.View ArticlePubMedGoogle Scholar
- Harada K, Fukusaki E, Bamba T, Sato F, Kobayashi A. In vivo 15N-enrichment of metabolites in suspension cultured cells and its application to metabolomics. Biotechnol Prog. 2006;22(4):1003–11.View ArticlePubMedGoogle Scholar
- Mesnard F, Ratcliffe RG. NMR analysis of plant nitrogen metabolism. Photosynth Res. 2005;83(2):163–80.View ArticlePubMedGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.