Rice Gene Coexpression

We performed gene co-expression analyses on all non TE-related genes across 80 libraries that are grouped into 32 tissues and 4 phases (seed, seedling, vegetative, and reproductive). We detected 39 co-expression modules. Module size ranged from 5 to 2,465 genes, with a median of 910 genes. We detected modules that exhibit the highest expression at each tissue. For example, we detected a module with peak expression at the embryo, which includes key morphogenic regulator BBM1 (PMID: 30542157) that plays a critical role in rice embryogenesis.

Download the complete gene coexpression module assignments for all rice genes:

osa1_r7.all_models.coexpression_modules.txt.gz - Gene coexpression module assignments for all gene models
• osa1_r7.all_models.coexpression_modules.xlsx - Gene coexpression module assignments for all gene models, Excel format
• osa1_r7.coexpression_module_list.txt.gz - List of all coexpression modules
• osa1_r7.coexpression_module_list.xlsx - List of all coexpression modules, Excel format

Rice Gene Coexpression Method:

Gene co-expression analysis was done according to Li et al., 2023 (PMID: 37063055). Gene expression (transcript per million, TPM) matrix was generated with Kallisto (PMID: 27043002). Gene flagged as transposable element associated genes were removed prior to analysis. TPM values were log-transformed, and biological replicates were averaged to calculate z-scores. Genes with no expression in any tissue were removed prior to downstream analyses. All pair-wise correlations were performed on the remaining 38569 genes. To build the correlation network edge table, only edges with r > 0.8 were used, which corresponded to the top 0.5% of all edges. A network object was constructed using the 'graph_from_data_frame()' function of igraph. Graph based clustering was performed using the Leiden algorithm (implemented as the 'cluster_leiden()' function in R as part of the igraph package) with a resolution parameter of 5.

This work is supported by grants (DBI-0321538/DBI-0834043) from the National Science Foundation and funds from the Georgia Research Alliance, Georgia Seed Development, and University of Georgia.