Dynamics of Gene Co-expression Networks in Time-Series Data: A Case Study in Drosophila melanogaster Embryogenesis

Abstract

Co-expression networks tightly coordinate the spatiotemporal patterns of gene expression unfolding during development. Due to the dynamic nature of developmental processes simply overlaying gene expression patterns onto static representations of coexpression networks may be misleading. Here, we aim to formally quantitate topological changes of co-expression networks during embryonic development using a publicly available Drosophila melanogaster transcriptome data set comprising 14 time points. We deployed a network approach which inferred 10 discrete co-expression networks by smoothly sliding along from early to late development using 5 consecutive time points per window. Such an approach allows changing network structure, including the presence of hubs, modules and other topological parameters to be quantitated. To explore the dynamic aspects of gene expression captured by our approach, we focused on regulator genes with apparent influence over particular aspects of development. Those key regulators were selected using a differential network algorithm to contrast the first 7 (early) with the last 7 (late) developmental time points. This assigns high scores to genes whose connectivity to abundant differentially expressed target genes has changed dramatically between states. We have produced a list of key regulators-some increasing (e.g., Tusp, slbo, Sidpn, DCAF12, and chinmo) and some decreasing (Rfx, bap, Hmx, Awh, and mld) connectivity during development-which reflects their role in different stages of embryogenesis. The networks we have constructed can be explored and interpreted within Cytoscape software and provide a new systems biology approach for the Drosophila research community to better visualize and interpret developmental regulation of gene expression.

Leveraging transcriptome and epigenome landscapes to infer regulatory networks during the onset of sexual maturation

“…The promise of multiomic data will remain unfulfilled without methodological approaches capable of identifying system perturbations associated with phenotypic change. Here, we used a gene regulatory network approach which has proven successful for investigation of puberty and other complex traits [ 20 , 65 , 66 ]. We identified TRIM25, a gene encoding the estrogen-responsive finger protein, as the network element undergoing the most dramatic change in connectivity and strongly suggests it plays a key role.…”

Transcriptome Response of Atlantic Salmon (Salmo salar) to a New Piscine Orthomyxovirus

“…A study in mice showed that animals lacking this gene rapidly succumbed to infection with influenza virus [ 34 ]. Overall, the significant changes in the gene co-expression networks between the HEALTHY and MORIBUND groups explored in this study suggest an important role of these transcription factors in the molecular events that underpin POMV infection [ 9 ].…”

“…A powerful approach to overcome this problem is gene co-expression network analysis [ 8 ]. Co-expression networks constructed from RNA expression data can effectively capture relationships between transcripts and reveal a biologically meaningful higher-order organization of the transcriptome [ 9 ].…”

Detection of selection signatures in farmed coho salmon (Oncorhynchus kisutch) using dense genome-wide information

“…Genomic scans for detecting selection signatures have been successfully applied to several domestic animals, including aquaculture species. For instance, selection signature scans have been carried out for Atlantic salmon 19 – 24 , Nile tilapia 25 , 26 ; and channel catfish 27 . To date, there are no studies focusing on the identification of selection signatures in farmed coho salmon populations.…”