Inferring gene regulatory network for cell reprogramming

The remarkable discovery of induced pluripotent stem cells (iPSCs) demonstrates that cell can be reprogrammed from somatic cell to a pluripotent state by the enforced expression of defined transcriptional factors. However, the underlying mechanism for cell reprogramming remains unknown and the regulatory interactions within this biological process have not been worked out. In particular from the gene regulatory network perspective, it is not clear how the four factors initialize the reprogramming process, propagate the information in a fine tuned way, and finally lead to the dramatic phenotype changes. In this paper, we analyze the time course gene expression data during cell reprogramming in mouse. We propose a three-stage procedure to infer gene regulatory networks. Specifically, we identify the major players during cell reprogramming by selecting differentially expressed genes in the first stage. Then in the second stage we utilize a new method to reveal strong correlations among those selected genes from short time series data. Finally the gene regulatory relationships are modeled by ordinary differential equations (ODE), the correlations are filtered by applying strong regularization, and directed and signed gene regulatory network for cell reprogramming is reconstructed. Preliminary analysis of the inferred network shows that short time series data provide biological insights for the dynamical process during reprogramming.