 Hello, my name is Maria and I would like to tell you about my graded project today. During the course of the project, we applied an unusual sampling scheme which drastically reduces the number of samples needed to study molecular circadian rhythm. I believe that it will be of interest to anyone who studies population genetics and who would like to optimize their experimental design. I will also briefly touch on atlas of circadian gene expression in fruit flies, which we created. Circadian rhythm is a biological process with a period of roughly 24 hours. It is spread on all levels of the organism from molecular rhythms inside the cells to behavior patterns. On molecular level, it consists of transcriptional and translational feedback loops which we in turn can observe as sinusoidal patterns of expression through time. The main question of our research is tissue specificity of a molecular circadian rhythm and genetic determinants thereof. We chose Drosophila melanogaster as an organism of study due to its simplicity of maintenance. It is also cheap and has no ethical concerns attached. In this presentation, I would like to shed a light on the shortcut to experimental design of a large-scale population genetic studies of molecular circadian rhythm. In order to understand why the shortcut is important, one needs to understand the scale of classical experimental setups in circadian biology. As an example application of classical scheme, I would like to demonstrate creation of atlas of tissue specific circadian transcriptome. For that, we took one commonly used white 1118 flyline and performed experiments under classical setup. We sampled this flyline every two hours for 48 hours in triplicate. As we were interested in tissue specificity of a circadian rhythm, we also sampled four major tissues, brain, gut, fat body, which is orthologous of human liver and malpigin tubules, which is orthologous of human kidneys. In total, we obtained 233 transcriptome, transcriptomes which is already quite a substantial amount. As a result, we could identify cycling circadian genes in every tissue and in total, we detected 1,729 cycling circadian genes across four tissues. Interestingly, the majority of the genes cycled in one tissue only, they were expressed in several tissues. However, that was a study of just one genotype. Imagine if we would like to study tissue specific genetics of a circadian rhythm under classical sampling scheme. In such case, we would need to sample a lot of genotypes. Usually this number starts around 150, multiplied by the number of tissue of interest in our case 3 and 12 time points as in classical scheme for sampling every two hours for 24 hours would result into astonishing 5,400 samples. It's not impossible, but it's a very laborious and financially exhausting task and this is where our shortcut experimental design comes in handy. Such sampling strategy was first implemented by Francesconi Lennon in application to development and I will demonstrate our adaptation of it to circadian rhythm on this animation. We will sample every genotype of interest just once. However, we will sample genotype very often. In our case, it will be every nine minutes. We call such samples static transcriptomes. Every dot on the video corresponds to a different genotype and you can see on the animation now that indeed sampling was done very, very frequently. Then, if technical noise is significantly smaller than biological signal, we will be able to recapitalate a sinusoid curve which you can see here of a circadian gene expression. In case a certain genotype results in the modified circadian rhythm, for example, with a shifted period or phase shift, we will detect such a line as an outlier of a curve as shown here with three dots. As a result, we will sample the same number of genotypes, 150, but we will sample them just once, therefore reducing the number of samples 36 times, meaning that we will have 150 samples per one tissue and that under 500 samples, we will be able to study all our free tissues of interest. All of this sounds great. However, the shortcut possesses several limitations one has to be aware about. First of all, since we are taking only one sample from each genotype, we will not be able to detect all genes which cycle with genotype and perform comparison of genotypes based on this metric. Next, our ability to catch a certain genotype which has a deviation in a period from normal 24 hours depends a bit on luck. This plot demonstrates a simulated curve of circadian gene expression, this colorful curve, recuperated from static transcriptome. The second curve formed by black triangles, this curve, shows an expression pattern of a genotype with an extended period of 30 hours in the imaginary case when we would sample it in a classical way. For demonstration purposes, I simulated sampling for 30 minutes rather than two hours for 24 hours. As you can notice, many points of the curve overlap each other, for example, like they do here or like they do here. It means that if we will take a static sample, according to our shortcut at worst time, for example, from 0 to 10 hours, the sample will fit the curve and we will not see it as an outlier. And in order to see it for an outlier it is, we would need to sample it in times indicated in these gray rectangles and these curves of expression are significantly different from each other. Of course, we can't know in advance which genotype will have an altered circadian clock, so we will have to be a bit lucky to sample a deviating genotype at a proper time point. The same situation as you can see takes place in case a genotype has a phase shift as shown on this slide. The triangles in this case show a phase shift of four hours with still normal period of 24 hours. As you could notice in both of these scenarios in a phase shift scenario as well as period change scenario, we will see both of these scenarios as an outliers and will not be able to distinguish between these different changes in the circadian rhythm. Unarguably, the advantages of the shortcut are significant reduction in number of needed samples and consequent reduction in labor and costs. All features of the shortcut could be summarized as following. It provides a binary response to the question if there is a change in circadian rhythm, just yes or no. However, after an outlier was identified, we can invest time and resources into performing a deeper study of that particular outlier using classical scheme and being sure that there is indeed a significant change in the molecular circadian rhythm. In our study, we applied the shortcut to a population of 140 genetically diverse fly lines and sampled three tissues from every line just once, resulting in still impressive number of 451 time-result transcriptomes. This slide shows that indeed sinusoidal curves, of expression of two major circadian genes, timeless and real, are recapitulated from static transcriptomes. You can even see an outlier circuit in black. Indeed, we found that 20% of sampled fly lines have tissue-specific circadian clocks deviating from expected biological time, which are displayed on this plot. The absolute time of deviation is reflected in the circle size and for some genotype it reached even 10 hours, for example, for this genotype 796. Translated to humans, such change in which circadian rhythm would be equal to a person waking up and feeling like nice and rested, but not hungry at all because his gut clock is 10 hours late. To sum up, I presented a shortcut to experimental design of large-scale population genetic studies, which will be particularly useful to study time-dependent biological processes, such as aging and development. I also outlined a resource of more than 700 tissue and time-result transcriptomes of fruit flies, which could further be used for other studies, for example, elucidating mechanism of tissue-specific gene expression. Finally, we believe that our Atlas of Circadian Gene Expression will also be useful for the circadian community. Last but not least, I would like to thank my colleagues who did an amazing job and without whom this study would not be possible. And thank you for your attention.