My team and I ended up primarily focusing on the first of the above-mentioned aims. Specifically we put a lot of effort into developing a computational simulation framework that allows its users to estimate power for a given study design (sample size, sampling time(s) and selection detection method) and a given epidemic. Using this framework, we then showed that most studies of selection driven by epidemics so far have been underpowered, including a study that was published in Nature in 2022. The latter was a result that ended up leading to us contributing to a rebuttal of that study (Barton*, Santander* et al, 2025). We also estimated that we had far too few samples to study several of the epidemics we originally planned to work on and how many samples we needed to look for at study the Black Death for instance. Finally, we made the tool freely available in the hope that future studies of selection driven by epidemics will be appropriately powered (Santander & Moltke, 2025).
We also took several first steps towards fulfilling the second aim, i.e. to develop a new and better method for detecting natural selection driven by epidemics. Specifically, we developed two new computational methods for detecting identity-by-descent (IBD), a feature in genetic data that we believe has great potential to be used for a new and better method for detecting selection. Both have resulted in an article and a freely available computational method (Nøhr et al, 2021 and Severson, Korneliussen & Moltke, 2022).
Finally, we also took several steps towards fulfilling the third aim. Specifically, we investigated real data relevant for three different epidemics: a severe Rinderpest epidemic in Cape Buffalo, a severe epidemic of kuru (a prion disease) in the 1900s among some of the communities in the Eastern Highlands of Papua New Guinea and the Black Death in European humans around 1349. The initial studies of the first two datasets led to papers about Cape Buffalo and a number of human communities in Eastern Highlands of Papua New Guinea, respectively (Quinn et al. 2023 and Quinn et al 2024). However, they also led to the conclusion that we did not have enough samples available to be able to detect selection with the data at hand for the rinderpest and the kuru epidemics. For the third epidemic, we initially reached the same conclusion. But near the end of the project, we finally managed to start a collaboration which led to the generation of a large enough dataset. We have started analysing this data and this has led to really interesting (but not yet published) results.
