ABSTRACT Most aspects of sleep and circadian rhythm are heritable, i.e., are in part determined by sequence variations in DNA. Multiple approaches are currently being applied to identify relevant genes influence sleep. One cutting- edge strategy being employed is high-diversity mouse models – particularly Diversity Outbred mice. These mice are derived through a novel randomized outbreeding approach seeded by the Collaborative Cross mice. This creates a genetically heterogeneous population of mice with genetic diversity approaching that in human populations. The genome of each Diversity Outbred mouse is a unique mosaic inherited from the original 8 Founder strains of mice. Accurate measurement of the genetic variation across the whole genome can be obtained using a specialized genotyping array. This data can then be combined with careful phenotyping of large numbers of Diversity Outbred mice to identify small quantitative trait loci containing only a few candidate causal genes. To utilize this resource to identify genes relevant for sleep, we have developed a high-throughput phenotyping pipeline that assesses multiple heritable aspects of sleep and circadian rhythm, including amounts of sleep, degree of sleep consolidation, vigilance, sleep drive, and circadian period. After discovering important loci using this high-throughput approach, robust validation of identified genes is carried out with gold-standard EEG/EMG recording of sleep in relevant strains of Collaborative Cross mice and, subsequently, in mice with a knockout of the predicted causal gene. Using this exact approach, we have already identified a novel gene regulating sleep in mice. Since this discovery, we have further increased the size of our phenotyped and genotyped sample of Diversity Outbred mice, and have now identified several other interesting quantitative trait loci containing candidate causal genes requiring validation. While existing analyses have shown promise in identifying important genes for sleep in mice, evidence from our work supports sleep as a complex genetic trait. That is, the phenotype is not simply determined by a single gene variant. Rather, there are likely to be complex genetic effects involving multiple interacting genes that determine additional variability in sleep/wake behavior. However, current analysis approaches in Diversity Outbred mice focus on additive genetic associations, and do not allow us to adequately address this concept. Thus, to uncover these more complex genetic effects, we propose to employ novel machine learning approaches to the wealth of available genetic and phenotypic data we have for hundreds of Diversity Outbred mice. Altogether, this is a high-risk, high-impact proposal appropriate for an R21 given the novel analytic strategy based on machine learning and the focus on validation of candidate causal genes affecting sleep. If successful, this proposal will provide new insights into the genetic underpinnings of sleep and a novel an...