Construction of time-delay gene network using coherence and cross spectrum evaluation

Analysis of gene expression data has become the base of gene network model construction. This facilitates the study of gene behavior in a collective manner. Many methodologies have been applied to construct gene network model. Most of these methods have not utilized the temporal aspects of gene expression data when there are time delays between gene pairs. A scheme has been put forward for time-delay gene network model construction by coherence and cross spectrum method in this paper. Gene expression data were viewed as a set of discrete time series from signal processing perspectives. Principles of power spectrum density and coherence function were introduced. Time-delay estimating method was discussed in detail. Gene network model for selected data set was constructed. The result demonstrates that our method can effectively reduce noise and detect time-delay relations between gene pairs at frequencies with high coherence.