Optimal control of probabilistic Boolean control networks: A scalable infinite horizon approach

• DOI: 10.1002/rnc.5909
• Publication Date: 2021-11-26T10:58:33Z
• Authors: Sonam Kharade; Sarang Sutavani; Sushama Wagh; Amol Yerudkar; Carmen Del Vecchio; Navdeep Singh
• Keywords: 0209 industrial biotechnology; 0206 medical engineering; 02 engineering and technology

Abstract One of the major issues in systems biology is developing control theory for gene regulatory networks (GRNs). Particularly, an important objective to develop therapeutic intervention strategies alter dynamics GRNs avoid undesired or diseased states. Several optimal have been developed find small (or medium) sized modeled as probabilistic Boolean (PBCNs). However, due humongous nature GRNs, we require strategy that scales large without posing any constraints on network dynamics. In this article, formulate infinite horizon discounted cost problem by leveraging Markov decision process (MDP) based PBCN structure model GRNs. Further, design a stationary strategy, thereby avoiding states associated with disease. An augmented state space defined facilitate method. By exploiting Kullback–Leibler (KL) divergence and log transformation function referred desirability function, devised using path integral (PI) approach. We propose sampling‐based technique approximation PI hence PBCNs. The amicable parallel implementation, addressing large‐scale Finally, viability results article shown some illustrative examples.