TEKRABber: A software for comparative analysis of gene regulatory networks including transposable elements

Author(s): Yao-Chung Chen,Arnaud Maupas,Katja Nowcik

Affiliation(s): Human Biology and Evolution, Institute of Bioinformatics, Freie Universität Berlin, Germany



Transposable elements (TEs), also known as “jumping genes”, are DNA fragments that can move within a genome and have paradoxically been seen as both a potentially deleterious genomic phenomenon and a potent driving force behind evolution. The genome-protecting KRAB zinc finger (KRAB-ZNF) proteins play a critical role in repressing TE expression within mammalian genomes, engaging in a dynamic interplay. This interplay was suggested to evolve according to an arms-race model, wherein TEs strive to transpose within the genome and KRAB-ZNFs adapt to suppress them. Despite indications of the involvement of TEs and KRAB-ZNFs in brain evolution and disease, a systematic evolutionary analysis across different primates is still lacking. Expression of genes between species can be compared by their orthology. However, due to the interspersed characteristic of TEs, the same TE subfamilies could be in different regions on chromosomes and have different sequence lengths. These make it difficult to compare their expression between species. Additionally, efficient tools are needed to estimate the pairwise correlations between orthologous genes and TEs for investigating the interplay between them.To efficiently compare the gene regulatory networks of KRAB-ZNFs and TEs across species, we developed an R Bioconductor software called TEKRABber. The name is derived from the idea that TEs are often repressed by KRAB-ZNF proteins. In a broader scope, TEKRABber addresses two primary challenges: comparing TE expression across species and efficiently calculating pairwise correlations between selected orthologous genes and TEs TEKRABber is designed to handle various types of transcriptomic read counts and offers two distinct modes of analysis. In the first mode, tailored for interspecies comparison, TEKRABber retrieves annotation files for orthologous genes and RepeatMasker annotations to estimate normalizing factors, ensuring comparable expression levels between species. Subsequently, users can employ the output data object to conduct differential expression analysis and identify one-to-one correlations based on selected parameters. Notably, the latest version of TEKRABber (version 1.8, Bioc3.19) includes a parallel computing option, significantly enhancing computational efficiency based on the number of cores a device can provide. The second mode is designed for comparing different conditions, such as control and disease states within the same species. In this scenario, users can bypass the normalization steps and directly generate data objects for conducting differential expression and correlation analyses. Furthermore, TEKRABber offers a user-friendly shiny dashboard function, providing users with an initial overview of their results before delving into the details. In our study, we used TEKRABber to explore the functional connections between KRAB-ZNFs and TEs in the context of human brain evolution and Alzheimer’s disease (AD). We conducted an analysis of KRAB-ZNF genes and TEs expression patterns and networks using two independent RNA-seq datasets: (1) Primate Brain Data: 33 human and multiple non-human primate brain regions, and (2) Mayo Data: temporal cortex and cerebellum of both healthy individuals and AD patients. Our analysis in Primate Brain Data highlighted species-specific expression variations, with many recently evolved TEs and KRAB-ZNF genes being differentially expressed between species, emphasizing their impact on evolution. Focusing on one-to-one correlations between TEs and KRAB-ZNF genes (TE:KRAB-ZNF), we found that KRAB-ZNFs significantly correlated more with TEs than other genes in human cortical regions. Additionally, comparing TE:KRAB-ZNF with published ChIP-exo data revealed many correlations involving recently evolved TEs and KRAB-ZNF genes that were specific to humans. From TE:KRAB-ZNF regulatory networks, we observed a clustering of young correlations around specific TE subfamilies such as Alu and SVA, shedding light on their influence on primate brain evolution. Integrating findings from the Mayo Data, we identified 14 pairs of TE:KRAB-ZNF uniquely detectable in the healthy human brain, suggesting dysregulating in AD brains. Our results deepen insights into primate brain evolution and offer a new perspective on human neurodegenerative disease through the analysis of TE:KRAB-ZNF regulatory networks. In summary, our research elucidates the intricate interplay between TEs and KRAB-ZNFs, unveiling their roles in shaping primate brain evolution and their dysregulation in AD. Through TEKRABber, we provide a pipeline tool for transcriptomic analysis, empowering developers and researchers to unravel the complexities of gene regulation including TEs across species and conditions.