Skip to main content
. 2020 Nov 6;82(4):1030–1046. doi: 10.1007/s00248-020-01596-5

Fig. 3.

Fig. 3

Random Forest classification model was applied on the 16S rRNA abundance data to identify most important microbial features in TT, NT, and saliva samples. Features with at least 4 reads and with a minimum prevalence of 10% across samples were included. Data was further transformed to centered log ratio (CLR) before applying the Random Forest classification algorithm