5. Conclusion & Future Work

5.1 Conclusion

This project employed time series analysis and machine learning tools to investigate whether Drosophila sleep have distinct stages and what characteristics define them. A new metric was developed on the basis of the five-minute immobility rule to identify sleep sessions from the velocity data (generated by the Ethoscope) of Drosophila activity over time. It was found that sleep during different parts of the day, namely mid-day, early-night and late-night, can be distinguished by a simple feature called burstiness. Specifically, early-night sleep that is thought to host deeper sleep stages exhibits the lowest burstiness, suggesting that burstiness may be a possible indicator for sleep intensity.

5.2 Future Work

Building from the exploratory analysis done in this project, further work can look into examining sleep stages with additional experimental manipulations. For example, arousal threshold experiments can be used to determine whether burstiness of activity during sleep is correlated with responsiveness to stimuli, as well as to identify other possible indicators of sleep intensity. In addition, fly behaviours can also be observed at a higher resolution to detect micromovements during sleep. This could be done by recording single fly placed in a circular arena. Flies can be tracked either using the same method as the Ethoscope or via a more refined method which recognises and tracks individual body parts. Overall, the ultimate goal is to identify measurable properties of different sleep stages in Drosophila and incorporate them into the Ethoscope software to facilitate future sleep-related studies.

Acknowledgements

I would like to thank L. Blackhurst for his guidance throughout this project and Dr. Gilestro for his helpful discussion and advice. I would also like to thank the Gilestro Lab members for sharing their research results and hosting journal clubs every week.