Fig. 6

Multi-omics combined analysis and ROC classification models establishment. A Protein–protein interaction network analysis illustrated the interconnections between 12 differentially expressed genes and proteins on the glycerophospholipid metabolism and phenylalanine metabolism pathways between NORA and CRA. Red represented 12 of the 40 differentially expressed genes between NORA and CRA, and blue represented the proteins on the glycerophospholipid metabolism and phenylalanine metabolism pathways. B Correlation heatmap and network revealed the interrelationship between differential bacteria, differential fungi, differential metabolites, and clinical inflammatory indicators, showing a strong correlation between flora, metabolites, and inflammatory features. Red represented the differential flora, purple represented the differential metabolites, and blue represented the clinical inflammatory features. The size of the graph represented degree, the thick line of the line represented the correlation, the solid line represented the positive correlation, and the dashed line represented the negative correlation. C, D ROC analysis demonstrated a combined AUC of 0.8519 and 0.8264 for training set and validation set of 4 features selected by logistic regression, respectively. E, F ROC analysis demonstrated a combined AUC of 0.8148 and 0.8056 for training set and validation set of 2 features selected by LASSO. G, H ROC analysis demonstrated a combined AUC of 0.9259 and 0.6736 for training set and validation set of 6 features selected by random forest. ROC: receiver operating characteristic