Structural Equation Modeling for Systems Biology

Inference of gene regulatory network is a useful approach to understand the control systems in living cells. To obtain the better insights into the gene regulation, we developed statistical approach, based on Structural Equation Modeling (SEM) in combination with factor analysis and improved four-step procedure. This approach allowed us to reconstruct a model of transcriptional regulation that involves protein-DNA interactions from only the gene expression data, in the absence of protein information. The significant features of SEM are the inclusion of latent variables within the constructed model and the ability to infer the network, including its cyclic structure. Furthermore, the SEM approach allows us to strictly evaluate the inferred model by using fitting scores. In this article, we’ll show the details of SEM approach for detecting the causality between genes and other cellular components.