*2.5. Gene Co-Expression Network Analysis*

Gene co-expression network analysis to group genes into modules used the R package WCGNA (v1.6.1). A power value of 6 and 9 for flag-leaf and panicle, respectively, predicted a gene co-expression network that exhibited scale-free topology with inherent modular features (Figure S10A,D). The "blockwiseModules" function of WGCNA was used to detect and generate modules. Network interconnectedness was measured by calculating the topological overlap using the TOMdist function with a signed TOMType. Average hierarchical clustering using the "hclust" function was performed to group the genes based on the topological overlap dissimilarity measure (1-TOM) of their connection strengths. Network modules were identified using the dynamic tree cut algorithm (version 1.63.1) with minimum and maximum module size of 20 and 20,000, respectively, merging threshold function at 0.15, and deep split parameter set at level 2. The module "eigengenes" was used to calculate the correlation coefficient for the samples to identify biologically significant modules. To visualize the expression profiles of the modules, the eigengene (first PC) for each module

was plotted using a customized bar plot function in R. To identify hub genes within the module, the module membership (MM) for each gene also known as module eigengene-based connectivity (kME) was calculated, based on the Pearson correlation between actual expression values and the module eigengene. Incorporating the gene significance (GS) measures, which could also be defined by the −log10 (*p*-value) from IL\_DvsSwa\_D contrast in DE analysis as external information into the co-expression network, genes within a module with the highest MM and GS values are highly connected within that module. To identify modules shared between the flag-leaf and panicle networks, a consensus network was generated. An in-house R function was used for overlap counting and statistical testing. The consensus network matrix for the flag-leaf and panicle tissue networks was plotted to show the significant overlap in gene count of two modules based on Fisher's exact test with the −log10 (*p-*value).
