Fig. 2

Hierarchical clustering and PCA analysis illustrate time, pretreatment and substrate dependant clustering patterns. With a hierarchical clustering of all conditions using the log transformed and quartile normalised mean FPKM values that were ≥ 1 in any of the time points on any media, b, c results from PCA showing separation in distinct clusters over the first 3 components in 2D and 3D respectively. Ellipses indicate 95% confidence intervals. Clusters that are discussed in the main text are labelled C1–C3 and indicated with coloured panels, with dashed lines representing sub-clusters. d Summarises relative abundance of GO-terms identified as linked to the clusters via word clouds