Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2015 Jun 11;3:87.
doi: 10.3389/fbioe.2015.00087. eCollection 2015.

MetabNet: An R Package for Metabolic Association Analysis of High-Resolution Metabolomics Data

Affiliations
Free PMC article

MetabNet: An R Package for Metabolic Association Analysis of High-Resolution Metabolomics Data

Karan Uppal et al. Front Bioeng Biotechnol. .
Free PMC article

Abstract

Liquid-chromatography high-resolution mass spectrometry provides capability to measure >40,000 ions derived from metabolites in biologic samples. This presents challenges to confirm identities of known chemicals and delineate potential metabolic pathway associations of unidentified chemicals. We provide an R package for metabolic network analysis, MetabNet, to perform targeted metabolome-wide association study of specific metabolites to facilitate detection of their related metabolic pathways and network structures.

Keywords: choline; metabolic networks; metabolic pathways; metabolite identification; metabolomics; targeted MWAS.

Figures

Figure 1
Figure 1
Overview of the study design.
Figure 2
Figure 2
Targeted MWAS for choline. (A) Type 1 Manhattan plot for Pearson correlation of targeted choline measurements (Metabolon) with each m/z feature in high-resolution metabolomics (HRM) for 50 plasma samples from common marmosets. (B) Comparison of Metabolon values with corresponding HRM analysis at Clinical Biomarkers Laboratory. Values are expressed as relative units. (C) Type 2 Manhattan plot of targeted choline measurements as a function of retention time of m/z feature suggests association of choline with lipids. (D) Type 3 Manhattan plot of targeted choline measurements as a function of ion intensity shows that correlated 12C and 13C forms of choline have expected differences in ion intensity and that other significantly associated and unidentified m/z features are present with lower ion intensities. (E) Internal Targeted MWAS is shown as type 2 Manhattan plot for Pearson correlation of choline within the HRM analysis. Results show significant associations of choline with m/z features eluting after 350 s. (F) Targeted MWAS for choline in human plasma. Type 2 Manhattan plot for Pearson correlation of choline within the HRM analysis of plasma of 50 healthy humans shows significant associations of choline m/z features eluting after 350 s. Mean choline concentration was 5.8 ± 0.8 μM. In (E) and (F), self-correlations of different forms of choline were removed to facilitate visualizations of correlation of choline with other m/z features. Positive correlations are shown in blue and negative correlations in red. FDR corrected significance level is shown as broken line.
Figure 3
Figure 3
Ions correlated with choline in both marmoset and human samples map to phosphocholine pathways in network enrichment analysis. (A) Venn diagram shows 67 ions were common among the 198 correlated ions for marmosets and 309 correlated ions for humans. (B) Analysis of these 67 ions by MetaCore showed significant pathway enrichments after FDR correction for multiple phosphocholine pathways. (C) Metabolome-wide network structure. Correlation analyses showed that 18 of the 67 ions had similar pairwise correlation patterns for marmosets and humans at p < 0.05. These were tested for correlations among other ions, and features that had absolute Spearman correlation >0.3 at FDR <0.05 are shown for marmosets and humans, demonstrating ability to map the broader network structure associated with choline. Stratification according to strengths of associations showed that the most highly associated clusters contained phospholipids, while terpenoids and steroids were present in less strongly associated clusters.
Figure 4
Figure 4
Systematic variation in stringency of correlation allows visualization of primary and secondary network structures. Analysis of ions positively and negatively correlated with a metabolite, as shown in Figure 2, aid in identification of unknown ions, especially when there are many database matches to the ion. By including secondary correlations in the analysis, as shown in Figure 3, one can begin to see a broader network organizational structure. This functionality is facilitated in MetabNet by providing ability to readily examine this network structure at different stringencies. At low stringency (|r| > 0.3), a large number of secondary correlations with choline are apparent. At greater stringency (|r| > 0.5), a number of secondary networks are evident in which many ions are significantly correlated with ions that are correlated with choline. This provides a basis to study perturbation in association with diet, genetic variation, and other causes of disease. Even higher stringency (|r| > 0.7) is useful in some cases to simplify the network structure to a small number of strong associations and also to facilitate identification of multiple ions, adducts, or isotopes derived from chemicals in the target list.

Similar articles

See all similar articles

Cited by 23 articles

See all "Cited by" articles

References

    1. Alonso A., Julià A., Beltran A., Vinaixa M., Díaz M., Ibañez L., et al. (2011). AStream: an R package for annotating LC/MS metabolomic data. Bioinformatics 27, 1339–1340.10.1093/bioinformatics/btr138 - DOI - PubMed
    1. Benjamini Y., Hochberg Y. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B Stat. Methodol. 57, 289–300.
    1. Chen Q., Park H. C., Goligorsky M. S., Chander P., Fischer S. M., Gross S. S. (2012). Untargeted plasma metabolite profiling reveals the broad systemic consequences of xanthine oxidoreductase inactivation in mice. PLoS ONE 7:e37149.10.1371/journal.pone.0037149 - DOI - PMC - PubMed
    1. Go Y. M., Walker D. I., Soltow Q. A., Uppal K., Wachtman L. M., Strobel F. H., et al. (2014). Metabolome-wide association study of phenylalanine in plasma of common marmosets. Amino Acids 47, 589–601.10.1007/s00726-014-1893-x - DOI - PMC - PubMed
    1. Ivanisevic J., Zhu Z. J., Plate L., Tautenhahn R., Chen S., O’Brien P. J., et al. (2013). Toward ‘omic scale metabolite profiling: a dual separation-mass spectrometry approach for coverage of lipid and central carbon metabolism. Anal. Chem. 85, 6876–6884.10.1021/ac401140h - DOI - PMC - PubMed

LinkOut - more resources

Feedback