Transcriptome-Wide Analysis of Gene Expression Landscape And Starch Synthesis Pathway Coexpression Network in Sorghum

Abstract

Background Gene expression landscape across different tissues and developmental stages reflects their biological functions and evolutionary patterns. Integrative and comprehensive analyses of all transcriptomic data in an organism are instrumental to obtaining a complete picture of their expression landscape and tissue specificity. Such studies are still very limited in an important crop plant, sorghum, which has been used as a popular model to study drought and temperature tolerance. It also limits the discovery of the genetic basis underlying complex agricultural traits in sorghum. Results We characterized the genome-wide expression landscape for sorghum using 873 RNA-sequencing (RNA-seq) datasets representing 19 tissues. Our integrative analysis of these RNA-seq data provides the most comprehensive transcriptomic atlas for sorghum, which will be valuable for the sorghum research community for functional characterizations of sorghum genes. Based on the transcriptome atlas, we identified 595 housekeeping genes (HKGs) and 2,080 tissue-specific express genes (TEGs) for the 19 tissues. We identified different gene features between HKGs and TEGs, and we found that HKGs have experienced stronger selective constraints than TEGs. Furthermore, we built a transcriptome-wide co-expression network (TW-CEN) comprising 35 modules with each model having specific Gene Ontology terms enriched. High connectivity genes in TW-CEN tend to express at high levels while undergoing intensive selective pressure. With the TW-CEN, we also built global and seed-preferential co-expression networks of starch synthesis pathways, which showed that several pathways such as photosynthesis, microtubule-based movement, carotenoid biosynthesis, glycerol ether metabolic processes play important roles in starch synthesis. Conclusions The global transcriptome atlas of sorghum generated by this study provides an important functional genomics resource for traits discovery and insight into starch synthesis regulation in sorghum.

An extended omnigenic model explains genome-phenome relationships for complex traits in global sorghum diversity

“…The gene loading for each trait was averaged over the marker loading within −5 kb of the gene (putative regulatory region) and gene body, and the marker loading was calculated using the rrBLUP using all phenotyped accessions 60 . The gene expression was extracted from a publicly available RNA-seq dataset across 19 tissues (https://zhenbinhu.shinyapps.io/Transcriptome) 33 . The correlation between gene expression and gene effect was calculated using the cor.test function with method = “spearman’’ in R.…”

“…Thus, for each trait and each gene, we averaged marker loadings from the GWP over the given gene, including exons, introns, and promoter region (5 kb upstream) to calculate the "gene loading". Then we calculated the relationship between gene loading for the biomass traits and RNAseq across 844 RNAseq data sets in 19 tissues 33 ( Fig. 1b ).…”