Structural and dynamical analysis of biological networks

Abstract

Biological networks are currently being studied with approaches derived from the mathematical and physical sciences. Their structural analysis enables to highlight nodes with special properties that have sometimes been correlated with the biological importance of a gene or a protein. However, biological networks are dynamic both on the evolutionary time-scale, and on the much shorter time-scale of physiological processes. There is therefore no unique network for a given cellular process, but potentially many realizations, each with different properties as a consequence of regulatory mechanisms. Such realizations provide snapshots of a same network in different conditions, enabling the study of condition-dependent structural properties. True dynamical analysis can be obtained through detailed mathematical modeling techniques that are not easily scalable to full network models.

CellOracle: Dissecting cell identity via network inference and in silico gene perturbation

“…Therefore, to interrogate the role of individual TFs within the network, we leveraged principles of graph theory to identify critical nodes, defined by different measures of centrality. For example, degree centrality is the most straightforward measure, reporting how many edges are connected to a node (in this case, how many genes a given TF connects to) (Klein et al, 2012). Here, degree centrality scoring in the MEP_0 cluster GRN configuration successfully recognized key TFs associated with erythrocyte differentiation; Gata1, Gata2, and Klf1, as above, in addition to Nfe2 (Andrews et al, 1993) and Ztbt7a (Norton et al, 2017) (Figure 3F; Figure S2B).…”

“…This analysis revealed that CellOracleinferred GRN configurations resemble a scale-free network, whose degree distribution follows a power law (Figure S2A). This configuration is characteristic of biological networks, defined by the presence of hub nodes serving as bridges between small degree nodes, supporting a robust network structure, in contrast to random networks (Klein et al, 2012). Many hub nodes are visible within CellOracle-inferred GRN configurations; for example, for the configuration inferred for the megakaryocyte erythrocyte progenitor (MEP) cluster, network modules and hubs can be distinguished ( Figure 3E).…”

In silico mutational studies of Hsp70 disclose sites with distinct functional attributes

“…For example, a close link between residue fluctuations (B‐factors) of a given protein and the average shortest path length to a residue from all others has been established . The network approach has enabled the study of specific proteins and has helped reveal interesting features not directly evident from the structure or sequence homology . For example, interaction conservation was utilized in phylogenetic analysis of remote homologs of the TIM barrel fold to reveal loop‐based conserved interactions near the active site .…”

Properties of Boolean dynamics by node classification using feedback loops in a network

“…For instance, the fate decision mechanism in a bacteriophage life cycle [1], the chemotaxis process in Escherichia coli [2], and segmental polarization in Drosophila melanogaster [3] were shown to be robust against noisy environments. It is more interesting that the dynamics of a biological network can be highly related to its structural characteristics [4]. In particular, many recent studies have shown that a feedback loop (FBL), a circular chain of interactions, can play an important role in controlling the robustness or susceptibility of networks [5, 6].…”