Figure 1.
Mummichog redefines the work flow of untargeted metabolomics.
A) In the work flow of untargeted metabolomics, the conventional approach requires the metabolites to be identified before pathway/network analysis, while mummichog (blue arrow) predicts functional activity bypassing metabolite identification. B) Each row of dots represent possible matches of metabolites from one m/z feature, red the true metabolite, gray the false matches. The conventional approach first requires the identification of metabolites before mapping them to the metabolic network. C) mummichog maps all possible metabolite matches to the network and looks for local enrichment, which reflects the true activity because the false matches will distribute randomly.
Figure 2.
Modular organization of human metabolic network.
A) Hierarchical clustering of the network by the steps between 4204 metabolic reactions, where the warmer color codes for fewer steps. Each red island represents a cluster of closely connected reactions. B) An insert by the while arrow in A. This demonstrates that network modules and pathways correlate with but not equate to each other. C) When measured by reaction steps between metabolites, most metabolites are connected in no more than four steps. This serves as a practical guide in searching subnetworks in the total metabolic network.
Figure 3.
Metabolic activity network in dendritic cells stimulated by yellow fever virus.
A) Prediction by mummichog directly from m/z feature tables (cell extracts after 6 hours of infection). Metabolites are colored according to log2 fold change. A high resolution copy is given in Figure S9. B) Further investigation was focused on the subnetwork on the right. Glutamate was not significantly altered, but included for network connectivity.
Figure 4.
Gene expression confirms the activity network.
A) Cytokines secreted after infection (ELISA) indicate the activation of innate immune programs. B) Down-regulation of transcripts of GCLC, GCLM (subunits of gamma-glutamylcysteine synthetase) and GSS (glutathione synthetase), the key enzymes for glutathione synthesis. C) Nitric oxide has feedback inhibition on the expression of eNOS and iNOS (nNOS was not detected). Gene expression was assayed by quantitative RT-PCR. Infected samples were compared to mocks by student's t-test (n = 3).
Figure 5.
Identification of metabolites by tandem mass spectrometry.
Arginine is shown as an example while the full data are given in Figure S3. From top to bottom: the fragmentation pattern ( followed by
on peak 157) from biological sample, from biological sample spiked with authentic chemical and from authentic chemical reference.
Figure 6.
Application of mummichog to additional data sets.
Metabolite prediction by mummichog is in good agreement with annotation in the original studies, 97% for the human urine data [63] and 86% for the yeast data [64]. The metabolites not in the original annotation (yellow) can not be compared. The “-z” option in mummichog enforces the presence of primary ion (M+H[+] for positive mode, M−H[−] for negative mode). This shifts the coverage in the huamn data set, but not much for the yeast data of limited annotation.
Table 1.
How metabolomics data match metabolic models.