Identification of biomarker in brain-specific gene regulatory network using structural controllability analysis

Background: Brain tumor research has been stapled for human health while brain network research is crucial for us to understand brain activity. Methods: Here the structural controllability theory is applied to study three human brain-specific gene regulatory networks, including forebrain gene regulatory network, hindbrain gene regulatory network and neuron associated cells cancer related gene regulatory network, hose nodes are neural genes and the edges represent the gene expression regulation among the genes. The nodes are classified into two classes: critical nodes and ordinary nodes, based on the change of the number of driver nodes upon its removal. Eight topological properties out-degree DO, in-degree DI, degree D, betweenness B, closeness CA, in-closeness CI, out-closeness CO and clustering coefficient CC) are calculated in this paper and the results prove that the critical genes have higher score of topological properties than the ordinary genes. Then two bioinformatic analysis are used to explore the biologic significance of the critical genes. Results: On the one hand, the enrichment scores in several kinds of gene databases are calculated and reveal that the critical nodes are richer in essential genes, cancer genes and the neuron related disease genes than the ordinary nodes, which indicates that the critical nodes may be the biomarker in brain-specific gene regulatory network. On the other hand, GO analysis and KEGG pathway analysis are applied on them and the results show that the critical genes mainly take part in 14 KEGG pathways that are transcriptional misregulation in cancer, pathways in cancer and so on, which indicates that the critical genes are related to the brain tumor. Finally, by deleting the edges or routines in the network, the robustness analysis of node classification is realized, and the robustness of node classification is proved. Conclusion: The comparison of neuron associated cells cancer related GRN and normal brain-specific GRNs (including forebrain and hindbrain GRN) shows that the neuron-related cell cancer-related gene regulatory network is more robust than other types.

Structural Controllability of Boolean Control Networks under Partial Information

It is worth noting that both nodes’ coupling connections and logical updating functions play a vital role in state evolutions of Boolean networks (BNs). In this paper, a new concept named structural controllability (SC) about Boolean control networks (BCNs) with known partial information on nodes’ connections is studied. Then, by referring to semi-tensor product (STP) techniques, several types of SC are presented according to different issues of Boolean functions. Thereafter, several necessary and sufficient conditions are derived for SC of BCNs. Finally, a biological model of the lactose operon in Escherichia coli is simulated to show the effectiveness of the main theoretical results.