Fig 1.
Metabox supports in-depth analysis of metabolomic data by including four analysis modules: data normalization (red), statistical analysis (blue), network construction (green) and functional analysis (purple). Outputs from each module are in red, blue, green and purple circles respectively. The tool accepts external inputs on each analysis level. Within metabox, the output from an analysis module can be used for subsequent analyses in the other modules denoted as a colored circle inside a box.
Table 1.
List of statistical analysis procedures in metabox.
Fig 2.
Partial visualization of the entire ‘chemical structure similarity’ network of metabolites in a lung adenocarcinoma study.
Chemical similarities between all identified metabolites was calculated from PubChem substructure fingerprints. Network nodes are connected by correlation coefficients using edge thickess for correlations rxy>0.7. Metabox functional class scoring was applied to estimate significantly enriched pathways (p<0.05), yielding arginine/proline metabolism, arginine biosynthesis and pentose/glucuronate interconversions among other pathways. Pathway enrichments are given by color in node pie charts.
Fig 3.
Biological network integrating significant differences in gene and metabolite regulation in lung adenocarcinoma compared to paired control tissues.
Significantly different genes and metabolites were mapped onto the metabox internal graph database using enzymes as linking nodes (grey). The resulting network was downloaded and mapped relative changes between tumor and non-tumor tissues using Cytoscape. Graph relationships CONTROL, CONVERSION, and CATALYSIS are labeled by colored edges. The network shows metabolic links between glycerol, palmitic acid, sphinganine, glutamic acid, glyceric acid, UDP-glucuronic acid and UDP-GlcNAc.
Table 2.
Comparison with existing tools for analysis and interpretation of metabolomic data.