TFmiR is a freely available web server for deep and integrative analysis of combinatorial regulatory interactions between transcription factors, microRNAs and target genes that are involved in disease pathogenesis. Since the inner workings of cells rely on the correct functioning of an enormously complex system of activating and repressing interactions that can be perturbed in many ways, TFmiR helps to better elucidate cellular mechanisms at the molecular level from a network perspective. The provided topological and functional analyses promote TFmiR as a reliable systems biology tool for researchers across the life science communities. TFmiR web server is accessible through the following URL: http://service.bioinformatik.uni-saarland.de/tfmir.

BackgroundIdentifying the gene regulatory networks governing the workings and identity of cells is one of the main challenges in understanding processes such as cellular differentiation, reprogramming or cancerogenesis. One particular challenge is to identify the main drivers and master regulatory genes that control such cell fate transitions. In this work, we reformulate this problem as the optimization problems of computing a Minimum Dominating Set and a Minimum Connected Dominating Set for directed graphs.ResultsBoth MDS and MCDS are applied to the well-studied gene regulatory networks of the model organisms E. coli and S. cerevisiae and to a pluripotency network for mouse embryonic stem cells. The results show that MCDS can capture most of the known key player genes identified so far in the model organisms. Moreover, this method suggests an additional small set of transcription factors as novel key players for governing the cell-specific gene regulatory network which can also be investigated with regard to diseases. To this aim, we investigated the ability of MCDS to define key drivers in breast cancer. The method identified many known drug targets as members of the MDS and MCDS.ConclusionsThis paper proposes a new method to identify key player genes in gene regulatory networks. The Java implementation of the heuristic algorithm explained in this paper is available as a Cytoscape plugin at http://apps.cytoscape.org/apps/mcds. The SageMath programs for solving integer linear programming formulations used in the paper are available at https://github.com/maryamNazarieh/KeyRegulatoryGenesand as supplementary material.Electronic supplementary materialThe online version of this article (doi:10.1186/s12918-016-0329-5) contains supplementary material, which is available to authorized users.

Putative disease-associated genes are often identified among those genes that are differentially expressed in disease and in normal conditions. This strategy typically yields thousands of genes. Gene prioritizing schemes boost the power of identifying the most promising disease-associated genes among such a set of candidates. We introduce here a novel system for prioritizing genes where a TF-miRNA co-regulatory network is constructed for the set of genes, while the ranks of the candidates are determined by topological and biological factors. For datasets on breast invasive carcinoma and liver hepatocellular carcinoma this novel prioritization technique identified a significant portion of known disease-associated genes and suggested new candidates which can be investigated later as putative disease-associated genes.

BackgroundSets of differentially expressed genes often contain driver genes that induce disease processes. However, various methods for identifying differentially expressed genes yield quite different results. Thus, we investigated whether this affects the identification of key players in regulatory networks derived by downstream analysis from lists of differentially expressed genes.ResultsWhile the overlap between the sets of significant differentially expressed genes determined by DESeq, edgeR, voom and VST was only 26% in liver hepatocellular carcinoma and 28% in breast invasive carcinoma, the topologies of the regulatory networks constructed using the TFmiR webserver for the different sets of differentially expressed genes were found to be highly consistent with respect to hub-degree nodes, minimum dominating set and minimum connected dominating set.ConclusionsThe findings suggest that key genes identified in regulatory networks derived by systematic analysis of differentially expressed genes may be a more robust basis for understanding diseases processes than simply inspecting the lists of differentially expressed genes.

Summary TFmiR2 is a freely available web server for constructing and analyzing integrated transcription factor (TF) and microRNA (miRNA) co-regulatory networks for human and mouse. TFmiR2 generates tissue- and biological process-specific networks for the set of deregulated genes and miRNAs provided by the user. Furthermore, the service can now identify key driver genes and miRNAs in the constructed networks by utilizing the graph theoretical concept of a minimum connected dominating set. These putative key players as well as the newly implemented four-node TF-miRNA motifs yield novel insights that may assist in developing new therapeutic approaches. Availability and implementation The TFmiR2 web server is available at http://service.bioinformatik.uni-saarland.de/tfmir2. Supplementary information Supplementary data are available at Bioinformatics online.