Fig. 5
From: Responsiveness to pulmonary rehabilitation in COPD is associated with changes in microbiota
Responders and non-responders to dyspnoea, exercise capacity and impact of the disease present distinct patterns of longitudinal bacteria-bacteria correlation. Correlation networks representing the longitudinal correlations between a subset of bacterial genera and ASVs: d0b698c7298bf04110a6d2f220879bfb and a5189f77a2cfeab3bc1602ff5c8ac3e9 non-responders and responders to A dyspnoea (rmcorr: [ 0.6–0.9]), B exercise capacity (rmcorr: [− 0.73–0.46]), and C impact of the disease (rmcorr: [− 0.71–0.44]). Grey triangles highlight common patterns of bacterial correlations between domains. Empty blue and red circles represent cytokines with which the correspondent taxon negatively and positively correlates, respectively. These correlations were inferred through the correlation analysis between bacteria-inflammatory markers and are represented here to facilitate the interpretation of the bacteria-bacteria correlation network. Prior to the analysis, genera and ASVs: d0b698c7298bf04110a6d2f220879bfb and a5189f77a2cfeab3bc1602ff5c8ac3e9 relative frequencies were transformed with arcsine square root transformation. The diameter of the nodes is proportional to the number of connections. Bacterial nodes were coloured according to the colour code of each bacterial phyla (Firmicutes: yellow, Bacteroidetes: blue; Proteobacteria: red; Fusobacteria: green Actinobacteria: pink; other phyla: grey). Positive correlations were represented in red, while negative correlations were represented in blue. The width of edges is proportional to the correlation coefficient