Random forest classification of pHTN, HTN and control using explanatory variables of CAGs + species. (a) ROC for the testing set consisted of controls, pHTN and HTN is performed based on the random forest model using the 1000 most important variables by ranking the variables by importance. The AUC is 0.67 for control versus pHTN (n = 12, red curve), AUC = 0.81 for control versus HTN (n = 12, green curve), and AUC is 0.47 for pHTN versus HTN (n = 13, blue curve). (b) The top 30 different CAGs or species distinguish HTN from control based on the random forest model. The bar lengths denote mean decrease accuracy, and the color represents CAGs or species enriched in control (blue), HTN (red), and neither (gray). (PDF 499 kb)