Skip to main content
Fig. 6 | Genome Biology

Fig. 6

From: Species-resolved sequencing of low-biomass or degraded microbiomes using 2bRAD-M

Fig. 6

Microbiome-based diagnosis of FFPE thin section from cervical cancer samples as enabled by 2bRAD-M. a Shannon and Simpson index among 15 healthy controls (H), 15 pre-invasive cancerous (PreC; benign), and 15 invasive cancerous (InvaC; malignant) samples. b Comparison of differential species and Lactobacillus spp. among the three groups. c The Random Forest classifier for discriminating cancer and healthy samples. In the ternary plot, each dot represents a FFPE sample. The axes indicate the microbiome-based probability of being InvaC, PreC, and H for a FFPE sample. The closer one sample is to an apex, the more likely it is predicted as to be corresponding disease states. d Feature selection by rebuilding Random Forest classifiers using a series of reduced sets of features. The scatter plot shows that nine variables (species) in a reduced RF model (i.e., the AUC plot on the right) can maximize model performance. And the ROC curve shows an even better discriminate performance using binary categories (averagely 0.96)

Back to article page