Fig. 2

Overview of iDMET for construction of differential metabolomic network based on the published metabolomic datasets. The procedure in iDMET consists of four main steps. The metabolomic data collected from papers or repositories are subjected to differential metabolomic analyses (step 1) to calculate the ratios of metabolite levels among the possible pairs of all conditions that appear in each paper. Step 1 can be omitted if the ratios are provided in the papers. Based on the specified thresholds of the ratios, up- and downregulated metabolites for each comparison of condition pair (designated as “differential metabolomic profile”) are selected (step 2). Then, for all possible pairs of comparisons, a 2 × 2 contingency table counting the number of metabolites that are upregulated (or downregulated) based on one differential analysis while upregulated (or downregulated) in the other differential analysis is generated. For each table, the odds ratio and p-value are calculated (step 3). Pairs of differential metabolomic profiles having remarkable odds ratios and p-values are selected and visualized as a network where each node represents a comparison (step 4). An edge between a pair of differential metabolomic profiles denotes that the upregulated (or downregulated) metabolites in one differential analysis significantly overlap with those that are upregulated (or downregulated) in the other differential analysis. Further details on the network visualization are given in the legend of Fig. 4