3.3 Regulatory genes evolution of the GSH metabolic pathway inG. hirsutum
In order to investigate the regulatory mechanisms of the duplicated GSH metabolic pathway in more detail, we constructed several co-expression networks based on 33 stress-related transcriptome data using the first 25,000 genes as targets (Fig. S1A). Cluster analysis was performed on the 33 transcriptome data to obtain the correlation coefficients based on the expression level of each gene.After calculating the weight values, the genes with high correlation were assigned to the same module, and the modules with similar expression were merged by dynamic tree-cut method (Fig. S1B). Then, we used the structural genes in the GSH pathway as the target genes to screen out all potential regulatory genes from the co-expression networks. Consequently, we found that under all stress conditions, only 21 structural genes (Table S5) appeared in these co-expression networks, while 98 regulatory genes were predicated to be correlated with the 21 genes ( Fig. S2, Table S5) .These regulatory genes were categorized into 13 gene families. Relatively abundant regulatory genes were identified from theWRKY (25), C2H2 (14), HSF (12), and ERF (10) gene families. In contrast, only two regulators were detected in four (B3, HD-Zip, NF-X1, and RAV) gene families (Fig. S3, Table S5).