Fig. 6: Use of Random Forest Regression Modeling to search for associations between immune-related peripheral blood gene signatures and changes in gastrointestinal microbiota and TTP.

The heatmap displays the sign of the derivative of the ALE curve (See Text). Blue/orange entries indicate features found to significantly associate with changes in a specific inflammatory pathway. Blue indicates a negative relationship, while orange a positive. For each immune pathway, a pathway-ASV association was determined significant if the Benjamini–Hochberg false discovery rate (FDR) adjusted p-value from the permutated importance analysis was found to be less than 0.05. Black symbols are used to identify the modeling-identified top important predictor (i.e., the predictor that if missing would lead to highest increase in mean squared error between model predictions and observations) for each specific host pathway. This analysis shows that reduction in TB burden and increased abundance of health-associated Cluster IV and XIVa Clostridia predicts inflammatory dampening. In contrast, increased abundance of oxygen-tolerant pathobionts including Enterococcus, Streptococcus, and E. coli predicts inflammatory exacerbation. A table reporting variable importance values, slope and intercepts from the ALE plot calculation and the related p values is provided in Supplementary Data 18. Source data are provided as a Source Data file.