Genetic Regulatory Network Modeling Using Network Component Analysis and Fuzzy Clustering

Gene regulatory network model is the most widely used mechanism to model and predict the behavior of living organisms. Network component analysis (NCA) as an emerging issue for uncovering hidden regulatory signals, has attracted significant trends in the research community. The common scheme in NCA is to model the controlling behavior of some proteins on the expression value of genes. However, this modeling requires performing certain experiments which are expensive in terms of time and feasibility. In this paper, we employ simple and effective data mining algorithm to obtain a purely gene- to gene model which predicts the effect of certain genes on the whole system. In order to accomplish this goal we employ fuzzy clustering and mutual information (MI) for determining regulator genes resulting in two methods named as: mutual information based NCA (MINCA) and fuzzy based NCA (FNCA). Simulation results validated using coefficient of determination (CoD), show that our methods model the system simpler and more accurate than conventional schemes