Figure 9.

Heatmap of species-wise standardized counts illustrating species module identification results (species are columns and modules are presented with different colors) using BZINB-based species spectral clustering. Each column represents a single species. Columns are ordered by the clusters predicted from the affinity matrix based on the BZINB correlations between species in the healthy (non-ECC) group. The columns are annotated to show and compare the estimated clusters within health (non-ECC) and disease (ECC) groups. Each row represents a participant, and the rows are ordered based on hierarchical clustering. Rows (n = 289) are annotated to denote healthy/disease and batch groups. Standardized counts were calculated separately for each species by first suppressing outliers (defined as greater than 3 × IQR from the first or third quartile), then centering around the median and dividing by the standard deviation.