Development of statistical and computational tools to analyze rhythmic gene expressions for a deeper understanding of circadian rhythms

Project Summary/Abstract Circadian rhythms are physiological, physical, and behavioral cycles that occur over a 24-hour period in living organisms. These rhythms are regulated by the rhythmic expression of specific genes. Key scientific questions revolve around how the rhythmicity of these genes is influenced by genotypes, feeding types, and drugs. More specifically, identifying genes that exhibit differential rhythmicity can help elucidate the effects of these factors. Current statistical tools and software can identify genes that show differential rhythmicity under various experimental conditions. However, in this high-dimensional context, there is a need for methods that are both computationally efficient and capable of accurately controlling the false discovery rate. Furthermore, there are currently no suitable methods to cluster differentially rhythmic genes or to map the gene networks among them. To address these needs, we propose three novel aims, each supported by innovative modeling or computa- tional algorithms. In Aim 1, we will develop a false discovery rate (FDR) controlled F-test with Welch’s correction to identify differentially rhythmic (DR) genes. In Aim 2, we will use a Dirichlet process mixture model with a novel computational technique to cluster these DR genes. In Aim 3, we will develop fast Bayesian algorithms to identify the gene networks. Once successful, these statistical methods and software will help reveal salient features of the rhythmic expression of genes, thereby benefiting circadian research, understanding the fundamentals of many cardio- vascular diseases associated with the disruption of circadian rhythms, and drug development.