WGCNA
We were firstly interested in which factors impacted the gene expression of Cladocopium and Durusdinium symbionts the most. To find this out we identified groups of genes (“modules”) that were co-regulated across samples using weighted gene co-expression network analysis (WGCNA). These modules, which WGCNA identifies in an unsupervised fashion without knowledge of the experimental design, are then examined for correlation with known traits or experimental treatments. In this way, it is possible to identify the largest and most responsive groups of genes and then investigate what biological effects they track. By examining the modules’ behavior across samples (Fig. 1) we observed that the greatest gene expression difference in the symbionts was not between “genus-background” and “genus-dominant” states, as we initially expected. Instead, in both genera the most distinct state was when the proportions of the two symbionts were near equal within the host (Fig. 1), the state that we here term “codominance.”
For Cladocopium, WGCNA identified nine modules of co-regulated genes. Two of these modules, containing 1233 and 131 genes respectively, were significantly correlated with codominance (r = 0.43, p < 4e-09 & r = 0.76, p < 2e-33). Additionally, the module containing 131 genes was negatively correlated with the dominance ofCladocopium (r = -0.26, p < 5e-04). ForDurusdinium , WGCNA identified 7 modules of co-regulated genes. Two of these modules, containing 815 and 81 genes respectively, were significantly correlated with codominance (r = 0.71, p < 9e-25 & r = 0.3, p < 8e-05). One module, containing 5671 genes was negatively correlated with codominance (r = -0.28, p < 2e-04). Notably, neither symbiont had modules significantly associated with exposure to elevated temperature.