Filtering of Data-Driven Gene Regulatory Networks Using Drosophila melanogaster as a Case Study

Abstract

Gene Regulatory Networks (GRNs) allow the study of regulation of gene expression of whole genomes. Among the most relevant advantages of using networks to depict this key process, there is the visual representation of large amounts of information and the application of graph theory to generate new knowledge. Nonetheless, despite the many uses of GRNs, it is still difficult and expensive to assign Transcription Factors (TFs) to the regulation of specific genes. ChIP-Seq allows the determination of TF Binding Sites (TFBSs) over whole genomes, but it is still an expensive technique that can only be applied one TF at a time and requires replicates to reduce its noise. Once TFBSs are determined, the assignment of each TF and its binding sites to the regulation of specific genes is not trivial, and it is often performed by carrying out site-specific experiments that are unfeasible to perform in all possible binding sites. Here, we addressed these relevant issues with a two-step methodology using Drosophila melanogaster as a case study. First, our protocol starts by gathering all transcription factor binding sites (TFBSs) determined with ChIP-Seq experiments available at ENCODE and FlyBase. Then each TFBS is used to assign TFs to the regulation of likely target genes based on the TFBS proximity to the transcription start site of all genes. In the final step, to try to select the most likely regulatory TF from those previously assigned to each gene, we employ GENIE3, a random forest-based method, and more than 9,000 RNA-seq experiments from D. melanogaster. Following, we employed known TF protein-protein interactions to estimate the feasibility of regulatory events in our filtered networks. Finally, we show how known interactions between co-regulatory TFs of each gene increase after the second step of our approach, and thus, the consistency of the TF-gene assignment. Also, we employed our methodology to create a network centered on the Drosophila melanogaster gene Hr96 to demonstrate the role of this transcription factor on mitochondrial gene regulation.

Dhr96[1] mutation and maternal tudor[1] mutation increase life span and reduce the beneficial effects of mifepristone in mated female Drosophila

“…Consistent with this relationship, Dhr96 functions specifically in the Drosophila midgut to promote lipid uptake and metabolism, and Dhr96[1] null mutant flies have reduced whole-body levels of TAG [ 33 , 34 ]. Recent network modeling using ChIP-Seq and protein-protein interaction data suggests that Dhr96 regulates TAG metabolism in part through regulation of mitochondrial genes to stimulate lipid consumption and mitochondrial respiration [ 45 ].…”

Transcriptome-derived networks reconstruct distinct immune strategies under different iron availability between Arabidopsis and rice

“…For GRNs, transcription factor-target associations were inferred with the GENIE3 package in RStudio using the DEGs under pathogen infection for each Fe condition and selecting the lists of Arabidopsis and rice transcription factors as regulators. GRNs were ltered to retain for each gene the 15 % most relevant transcription factors and thus avoid spurious connections adapted as from previously described 77 . The Igraph package in RStudio was used to assess topological information of networks and community composition according to the FastGreedy algorithm.…”