A Structural Output Controllability Approach to Drug Efficacy Prediction

Recent progresses in research and development of directed human protein-protein interaction (PPI) networks have opened a great opportunity for the uncovering of possible effects that a drug has during the drug therapeutic processes. Such PPI networks can be used to help researchers predict cell signaling pathways and gain better understanding about drug healing and targeting mechanisms. This paper aims at examining the use of network theory approach to study and analyze the dynamics and treatment of disease in samples of PPI networks. In this paper's analysis, the so-called disease nodes are defined as nodes that are responsible for the spreading of the considered disease. By combining the concept of network output controllability with an approach for network simplification, this paper shows that the number of controllable disease nodes for a particular disease can be predicted using a relatively simple computational method. Furthermore, the analysis reported in this paper indicates that a positive linear relationship between the number of controllable disease nodes and the relative efficacy of the used drug is statistically significant in term of drug targeting scheme. While the availability of directed human PPI networks is currently remains incomplete, the results reported in this paper can further corroborate the advantages of using concepts and techniques from control theory and network science in the study and research on drug discovery.