Somnath Datta. Exploratory Statistical Analysis of Differential Network Behaviors Based on Gene Expression Atlas of Palate Development

An extensive gene expression atlas for craniofacial development including that of the murine palate in mice has been constructed and validated by the research team of Steven Potter as part of the FaceBase Consortium. Since genes act in consort during a biological process, a network analysis is essential for a system-wide understanding. To the best of our knowledge, such a system-wide network and differential network analysis has not been conducted in the literature for the murine palate development. The goal of this R03 proposal is to mine these data for the E14.5 palate to explore further into the collective workings of the genes for various compartments at this stage. This will be achieved by reverse engineering of gene-gene association/interaction networks and their differential analysis. In the process, exiting and novel modern statistical techniques will be employed and tools (R packages) will be developed that will be freely available to the scientific community for other similar studies involving microarray and next generation sequencing based data collected across multiple experimental conditions. Thus, the following two broad interconnected aims (and sub-aims) all leading to better understanding of craniofacial palate development in mice will be undertaken. We will explore suitable measures for detecting gene-gene association/interaction using RNA-seq expression data and construct gene-gene association/interaction networks for palate development using these association measures (Aim 1). We will conduct a differential network analysis based on the gene expressions between two regions; this will show how the connected components vary in different regions of the palate indicating changes in collective activities of the corresponding genes (Aim 2). Also, we will identify genes paying important cross compartment roles by integrating ranked lists of differentially connected genes between multiple pairwise comparisons. It is also anticipated that the proposed research will lead to some novel bioinformatics tools (R packages/codes) which will be freely distributed through the Comprehensive R Archive Network and/or Bioconductor.