Network models are the focus of a growing number of researchers concerned with discovering novel gene interactions and regulatory relationships between genes from expression data. In this talk we will discuss two approaches for inferring networks from time-course expression data. In the first half of the talk we will present a model-based approach that unifies the processes of inferring networks and clustering genes. Specifically, we provide a probabilistic framework for inferring clusters from gene expression profiles. Genes within the same cluster are expected to share a similar expression profile. We build a network over clusters using state-space models. In the second half of the talk, we will discuss an approach for inferring networks from time course microarray data which relies on modeling gene expression profiles as random functional transformations of a reference curve. Using measures of functional similarity and time order based on estimated warping functions we discuss time-varying networks. We will illustrate the methods with simulation studies and a case study using time course microarray data arising from animal models on prostate cancer progression.
This is joint work with D. Telesca, M.Neira, C. Nelson, M.Gleave and R. Etzioni.
