Identification of Biologically Essential Nodes via Determinative Power in Logical Models of Cellular Processes

Authors

Trevor Pentzien, University of Nebraska at Omaha
Bhanwar L. Puniya, University of Nebraska-LincolnFollow
Tomas Helikar, University of Nebraska-LincolnFollow
Mihaela T. Matache, University of Nebraska at OmahaFollow

Document Type

Article

Date of this Version

8-31-2018

Citation

Frontiers in Physiology | www.frontiersin.org August 2018 | Volume 9 | Article 1185 Page 1-20

Comments

Pentzien T, Puniya BL, Helikar T and Matache MT (2018) Identification of Biologically Essential Nodes via Determinative Power in Logical Models of Cellular Processes. Front. Physiol. 9:1185. doi: 10.3389/fphys.2018.01185

Abstract

A variety of biological networks can bemodeled as logical or Boolean networks. However, a simplification of the reality to binary states of the nodes does not ease the difficulty of analyzing the dynamics of large, complex networks, such as signal transduction networks, due to the exponential dependence of the state space on the number of nodes. This paper considers a recently introduced method for finding a fairly small subnetwork, representing a collection of nodes that determine the states of most other nodes with a reasonable level of entropy. The subnetwork contains the most determinative nodes that yield the highest information gain. One of the goals of this paper is to propose an algorithm for finding a suitable subnetwork size. The information gain is quantified by the so-called determinative power of the nodes, which is obtained via the mutual information, a concept originating in information theory. We find the most determinative nodes for 36 network models available in the online database Cell Collective (http:// cellcollective.org). We provide statistical information that indicates a weak correlation between the subnetwork size and other variables, such as network size, or maximum and average determinative power of nodes. We observe that the proportion represented by the subnetwork in comparison to the whole network shows a weak tendency to decrease for larger networks. The determinative power of nodes is weakly correlated to the number of outputs of a node, and it appears to be independent of other topological measures such as closeness or betweenness centrality. Once the subnetwork of the most determinative nodes is identified, we generate a biological function analysis of its nodes for some of the 36 networks. The analysis shows that a large fraction of the most determinative nodes are essential and involved in crucial biological functions. The biological pathway analysis of the most determinative nodes shows that they are involved in important disease pathways.