Fig. 2

Some results of IP4M using a real world data set. ‘G1’ indicates the group of 1-week-old mice and ‘G2’ indicates the group of 7-week-old mice. a, b BPCs of LC–MS and TICs of GC–MS data sets derived from murine brain metabolic profiles. c–h Some results of multivariate differential analysis: PCA scores plot (c); OPLS-DA scores plot (d); the top 10 potential markers provided by RF (e) and SVM (f); the venn plot (g) of potential markers from RF, SVM and Boruta; and the validation boxplot (h) of Boruta based on the union set of the top 10 markers from RF and SVM. i, j Results of the correlation analysis: r value heatmaps GRaMM (i); scatter plot with nonlinear fitting curve of the correlated pair “DHA ~ Firmicutes” derived from GRaMM (j). k–n Some results of pathway analysis: the bubble plot of pathway analysis using “relative—betweenness centrality” algorithm (k) and “eigenvector centrality” algorithm (l). Each bubble corresponds to a metabolic pathway. The x-coordinate indicates the extent of pathway influence (Pathway Impact). The point size is related to Pathway Impact of the pathway. The ordinate represents the negative logarithm of the p value obtained from the enrichment analysis. Pathways with p < 0.05 are labeled. m, n The network diagrams of two significant pathways. Up and downregulated metabolites (G2 versus G1) are highlighted in red and blue respectively