Spearman correlations (ρ) between stool metabolite concentrations with the relative abundance of five genera more abundant in ABIS controls (Faecalibacterium, Bifidobacterium, Roseburia, Akkermansia, and Coprococcus). Stool metabolites were identified using machine learning with the random forest regressor in Python to extract the top 20 metabolites predicting the relative abundance of the respective genus, followed by Spearman correlations. Metabolites with significant (p < 0.05) correlations are noted with the coefficient. Instances within plot of multiple metabolites with the same name demonstrate adducts with distinct m/z ratios and retention time (generally the loss of H2O).
