Fig. 7

Metabolites, transcripts, and transcript/gene ratios can differentiate IBS-D from IBS-C with high accuracy. A ROC curves for random forest classifiers constructed from differentially abundant predicted genes, transcripts, and metabolites (colored in red) are compared to the ROC curve for a multi-omics classifier (colored in blue). The multi-omics classifier was constructed from significantly upregulated transcripts (by RNA/DNA ratio) in enriched pathways and differential metabolites associated with microbial community metabolic potential. All classifiers were trained on 60% of the dataset and tested on the remaining 40% of samples (n = 126 with all three data types). Colored areas indicate the 95% confidence intervals of the ROC curves. P values were calculated by bootstrapping. B–D Importance scores are shown for features included in the B metatranscriptomics, C metabolomics, and D multi-omics classifiers, with bar color denoting features that were enriched or depleted in IBS-D