Three topological features of regulatory networks control life-essential and specialized subsystems

Abstract

Gene regulatory networks (GRNs) play key roles in development, phenotype plasticity, and evolution. Although graph theory has been used to explore GRNs, associations amongst topological features, transcription factors (TFs), and systems essentiality are poorly understood. Here we sought the relationship amongst the main GRN topological features that influence the control of essential and specific subsystems. We found that the Knn, page rank, and degree are the most relevant GRN features: the ones are conserved along the evolution and are also relevant in pluripotent cells. Interestingly, life-essential subsystems are governed mainly by TFs with intermediary Knn and high page rank or degree, whereas specialized subsystems are mainly regulated by TFs with low Knn. Hence, we suggest that the high probability of TFs be toured by a random signal, and the high probability of the signal propagation to target genes ensures the life-essential subsystems’ robustness. Gene/genome duplication is the main evolutionary process to rise Knn as the most relevant feature. Herein, we shed light on unexplored topological GRN features to assess how they are related to subsystems and how the duplications shaped the regulatory systems along the evolution. The classification model generated can be found here: https://github.com/ivanrwolf/NoC/.

Topological benchmarking of algorithms to infer gene regulatory networks from single-cell RNA-seq data

“…For this reason, here we compute four centrality measures that have been previously adopted to find hubs in GRNs ( Koschützki and Schreiber 2008 ): the Betweenness ( Freeman 1977 ), the Centrality ( Freeman 1978 ), the Radiality ( Valente and Foreman 1998 ), and the Page Rank ( Page et al 1999 ) (see Methods section). Among these, the Page Rank and the Centrality metrics are conserved along evolution and relevant in pluripotent cells ( Wolf et al 2021 ). Moreover, they were proposed as metrics to distinguish life-essential versus specialized subsystems ( Wolf et al 2021 ).…”

Review and assessment of Boolean approaches for inference of gene regulatory networks

“…In the context of GRN inference, this might prove to be problematic for two reasons. Firstly, average connectivity within a network and thus prevalence of edges can differ significantly among different species [67] . Secondly, if we presume that the average connectivity is independent of a (sub)network size within the same species, the number of edges within the network will grow linearly with the network size.…”

An updated gene regulatory network reconstruction of multidrug-resistant Pseudomonas aeruginosa CCBH4851

“…Studies have shown a scale-free structure in cellular metabolic networks, 32 , 40 protein interaction networks, including in cancer, 41 , 42 transcription regulatory networks, and GRN. 20 , 43 , 44 , 45 Following the literature, 36 , 37 , 46 , 47 , 48 there are some qualitative and quantitative characteristics to ensure that a network is scale-free: the power-law distribution appears as a straight line on a log-log plot; the γ value usually is in the range 2<γ<3; and the presence of high-degree nodes, called hubs, the most highly connected nodes, 47 with most nodes clustered around them. The hubs demonstrate the absence of a uniform connectivity distribution in the network, presenting the 80-20 rule (also referred to as the Pareto principle), with small-degree nodes being the most abundant.…”