Network Control and Damage Mitigation in Complex Networked System

Open Access
Author:
Yang, Gang
Graduate Program:
Physics
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
December 14, 2017
Committee Members:
  • Réka Albert, Dissertation Advisor
  • Réka Albert, Committee Chair
  • Dezhe Jin, Committee Member
  • Tim Reluga, Committee Member
  • Lingzhou Xue, Outside Member
Keywords:
  • Network Control
  • Network Dynamics
  • Boolean Network
  • Damage Mitigation
  • Logical Modelling
Abstract:
Dynamical models have been successfully employed to study how different molecular components give rise to to cellular functions in biological systems. Such models are of great importance in understanding the underlying mechanisms of complex disease and designing preventive or therapeutic strategies. Complex diseases often start with abnormal mutations of the system, which can be modeled as network damage in network dynamical model. Network control problems design strategy to influence or drive the system to a desired state. Thus network control and damage mitigation strategies are a promising avenue toward developing disease intervention and therapies. Developing network dynamical models and network control strategies are challenging tasks as numerous components interact in a diverse ways in a biological system and we often have incomplete information including the dynamical mechanism and precise quantitative parameters from experimental data. Logic dynamical models, such as Boolean network models, demonstrate their value in such situations, including having considerable dynamic richness and capacity to capture emergent characteristic of real biological systems. I contributed to developing two framework to solve network control problem in Boolean network models. I design compensatory interactions to mitigate multiple network deregulations and stabilize the system as disease prevention or immediate treatment method. I also applied a heuristic algorithm to solve the target control problem in Boolean network models, which can be used to design disease treatment. This heuristic algorithm is based on a concept called domain of influence of node states, which describes the stabilization effect of a long-term intervention. These two frameworks complement each other in their method and purpose. Another way to proceed with incomplete information is structure-based control of continuous models. I test and compare two established methods, structural controllability and feedback-vertex control to understand their differences and elucidate relationship between network topology and network dynamics. I applied the above framework to several real biological models, including dynamical models involved in complex disease such as cancer and T-LGL leukemia. These analytical and computational tools not only generate solutions consistent with established experimental results and previous established tools, but also make predictions to help guide experimentalist to design real solutions to the challenging complex diseases. These developed frameworks in my dissertation also points out new directions for future research in network control. The dissertation is organized in the following way. In chapter 1, I introduce the background, concepts and methods involved in network modeling and network control problems. In chapter 2, I report the work to design immediate damage mitigation strategies through compensatory interactions in Boolean network models. In chapter 3, I report the work to solve the target control problem in Boolean network models through the concept of domain of influence. In chapter 4, I applied the structure-based network control strategies for continuous models to real biological systems. In chapter 5, I discuss possible future works and some preliminary results, especially those inspired by combining ideas from multiple previous chapters.