A Structural Characterisation of the Mitogen-Activated Protein Kinase Network in Cancer

Abstract

Gene regulatory networks represent collections of regulators that interact with each other and with other molecules to govern gene expression. Biological signalling networks model how signals are transmitted and how activities are coordinated in the cell. The study of the structure of such networks in complex diseases such as cancer can provide insights into how they function, and consequently, suggest suitable treatment approaches. Here, we explored such topological characteristics in the example of a mitogen-activated protein kinase (MAPK) signalling network derived from published studies in cancer. We employed well-established techniques to conduct network analyses, and collected information on gene function as obtained from large-scale public databases. This allowed us to map topological and functional relationships, and build hypotheses on this network’s functional consequences. In particular, we find that the topology of this MAPK network is highly non-random, modular and robust. Moreover, analysis of the network’s structure indicates the presence of organisational features of cancer hallmarks, expressed in an asymmetrical manner across communities of the network. Finally, our results indicate that the organisation of this network renders it problematic to use treatment approaches that focus on a single target. Our analysis suggests that multi-target attacks in a well-orchestrated manner are required to alter how the network functions. Overall, we propose that complex network analyses combined with pharmacological insights will help inform on future treatment strategies, exploiting structural vulnerabilities of signalling and regulatory networks in cancer.

Deep Reinforcement Learning for Stabilization of Large-scale Probabilistic Boolean Networks

“…The connectivity of the PBN with 20 nodes is as follows: inp 1 = [3, 6], inp 2 = [7, 14], inp 3 = [3, 5], inp 4 = [7, 4], inp 5 = [9, 6], inp 6 = [3, 11], inp 7 = [11, 3], inp 8 = [10, 9], inp 9 = [14, 7], inp 10 = [8, 19], inp 11 = [8, 6], inp 12 = [9, 4], inp 13 = [14, 16], inp 14 = [14, 18], inp 15 = [19, 15], inp 16 = [19, 2], inp 17 = [18, 4], inp 18 = [1, 20], inp 19 = [2, 5], inp 20 = [18, 20], where, again, inp i denotes the set of nodes that provide input to node i .…”

Deep Reinforcement Learning for Stabilization of Large-Scale Probabilistic Boolean Networks

“…In this context, intervention at a certain time takes the form of effecting a perturbation on the state of a node, which in GRNs translates to knocking-out a gene (switch to 0) or activating it (to 1). A lot of changes in cancer are of a regulatory epigenetic nature [13] and thus cells are expected to be re-configured to achieve the same outcome, e.g., inputs to gene regulatory elements by one signalling pathway can be substituted by another signalling pathway [14]. Hence, cancer biology suggests that perturbations may be transient, e.g., see single-step perturbations in [15].…”