*2.8. Identifition of Weighted Gene Co-Expression Network*

To obtain candidate key genes associated with the phenotypical traits, WGCNA [20] was used to distinguish the specific genes that are related to the traits, including tiller number, N content in leaf and in stem, dry weight of root, stem and leaf etc. After removing the genes with low FPKM levels, 21,700 genes were retained for the further WGCNAs analysis. This analysis identified twenty-two distinct gene co-expression modules (labeled in different colors) shown in the dendrogram (Figure 7A), and a total of eleven modules (brown, green, darkred, red, blue, magenta, pink, tan, lightyellow, turquoise and purple) were significantly related to the above-mentioned various phenotypical traits, respectively (Figure 7B). The eigengene expression in blue and green modules (M10 and M3) are shown in Figure 7C,D, respectively. The tiller number was positively most correlated with the eigengene expression levels of the lightyellow module (M18), and the correlation coefficient (*r*) was 0.61 (*p* = 0.04). Gene ontology (GO) annotation analysis reveals that the cellular components of this module genes mostly fall into extracellular region, yet molecular functions and biological processes are not significantly enriched (Table S7). The tiller number was negatively most correlated with the green module, with *r* = −0.71 (*p* = 0.01). GO annotation shows that the biological processes of the green module genes fall into biological regulation, regulation of cellular process, transport, establishment of localization, localization, regulation of transcription, etc. The molecular function fall into transporter activity, protein binding, transmembrane transporter activity, active transmembrane transporter activity, and transcription regulator activity. The cellular component of this module fall into membrane, integral to membrane, intrinsic to membrane, and cell part (Table S8).

**Figure 7.** Weighted gene co-expression network analysis (WGCNAs) of gene expressions and the related traits. (**A**) Hierarchical cluster tree showing 22 co-expression modules identified by WGCNAs. Each branch in the tree represents an individual gene; (**B**) Matrix of module – trait correlation: A total of 22 modules shown on the left and the relevance color scale from −1 to 1 is on the right. The numbers in parentheses represent the significance (*p*), and the numbers above represent the correlation coefficient (*r*). The eigengene expression profile in blue module (**C**) and green module (**D**).

The scales of N content in stem and in leaf were highly positively correlated with the eigengene expression in the blue module (M10), with *r* at 0.71 (*p* = 0.01) and 0.69 (*p* = 0.01), respectively. GO annotation shows that the biological processes of the blue module genes fall into the translation, gene expression, cellular protein metabolic process, small molecule metabolic process, protein folding, protein metabolic process, cellular ketone metabolic process, carboxylic acid metabolic process etc. The molecular functions of the blue module are the structural constituent of ribosome, structural molecule activity, translation factor activity, nucleic acid binding, pyrophosphatase activity, and GTP binding. Their cellular components are cytoplasm, macromolecular and ribonucleoprotein complex, ribosome, and cytoplasmic part (Table S9). And the scale of N content in stem and in leaf were all negatively correlated with the green module (*r* = −0.81, *p* = 0.002) and (*r* = −0.74, *p* = 0.006), consistent with the negative correlation module of the scale of tillers. In addition, the modules related to other traits and their GO annotations are shown in Figure 7B and Tables S10–S14. Interestingly, modules positively related with N content are in a negatively association with phosphorus content, and vice versa for the positively related modules. Their associations with sulphur, calcium and potassium content are in a similar pattern but not that consistent. These reveal that the phenotypic traits are more closely associated with certain modules of genes than other modules.
