Now showing 1 - 1 of 1
- ItemGenealogical properties of spatial models in Population Genetics(Hannover : Technische Informationsbibliothek, 2023-09) Wirtz, JohannesAt the interface between Phylo- and Population Genetics, and recently heavily inspired by Epidemonology, the discipline of Phylogeography comprises modelling techniques from classical theoretical biology and combines them with a spatial (2D or 3D) aspect, with the purpose of utilizing geographical information in the analysis to understand the evolutionary history of a biological system or aspects of virology such as directionality and seasonality in pandemic outbreaks [1, 2, 3, 4]. An prime example of this are datasets that take into account the sampling locations of its components (geo-referenced genomic data). In this project, we have focused on the model called "spatial Lambda-Fleming-Viot process" ( V [5, 6]) and analzed its statistical properties forward in time as well as in the ancestral (dual) process, with results that may be used for parameter inference. Of particlar interest was the spatial variance, denoted , a parameter controlling the speed at which genetic information is spread across space and therefore an analog of the reproduction number (R0) used in epidemonology e.g. to assess the infectiousness of differing viral strains. We explored the relation of this parameter to the time to coalescence between lineage pairs in this model and described methods of estimating it from sampled data under different circumstances. We have furthermore investigated similarities and differences between this model and classical models in Population Genetics, particularly Birth-Death processes, which are heavily used for all kinds of biological inference problems, but do not by themselves feature a spatial component. We compared the Vto a variant of the Birth-Death process where the location of a live individual changes over the course of its lifetime according to a Brownian motion. This process is not as easily viewed backward in time as the V, but the genalogical process is accessible by Markov-Chain Monte Carlosimulation, as the likelihoods of ancestral positions and branch lengths are easily calculated, making this model easily applicable to data. Our analysis highlights the analogy between the two processes forward in time as well as backward in time; on the other hand, we also observed a divergent behavior of the two models when no prior on the phylogenetic time scale was assumed. Lastly, this project has given rise to a study of combinatorial properties of tree shapes relevant to the V, the Birth-Death and other biological processes. In particular, we were able to identify the combinatorial class genealogical trees generated from these processes belong to and verify a conjecture regarding their enumeration. Preliminary versions of software tools for the aforementioned inference have also been provided.