Inferring the genetic basis of sleep states in Drosophila melanogaster using hidden Markov models

Abstract Multiple lines of evidence suggest that sleep in flies is not a unitary state but has lighter and deeper stages similar to mammals. A hidden Markov model applied to activity count data revealed lighter and deeper sleep states in a wild-derived population of flies. Fitting the data to a range of possible sleep states enabled us to determine the optimal number of sleep states for each fly. Four sleep and waking states were optimal across the population. We verified physiological differences among states experimentally using an arousal threshold paradigm. We then calculated the time spent in each sleep state for each fly. The time spent in each sleep state was heritable and therefore mappable to the genome. We mapped these parameters to the genome, identifying state-specific genes. Additionally, we provide software, FlyDreamR , that users can apply to activity count data.