Time-sensitive reconstruction of joint evolutionary trees in a host-parasite model with host switches

by Zsuzsanna Bösze
Doctoral thesis
Date of Examination:2024-10-22
Date of issue:2024-12-20
Advisor:Prof. Dr. Anja Sturm
Referee:Prof. Dr. Anja Sturm
Referee:Prof. Dr. Christoph Bleidorn
Persistent Address: http://dx.doi.org/10.53846/goediss-10973

 

 

Files in this item

Name:Bosze_thesis_rev.pdf
Size:2.91Mb
Format:PDF

The following license files are associated with this item:

Abstract

English

This thesis focuses on mathematical models that describe the joint evolution of host-parasite systems in the presence of host switches, or in other words, horizontal transfers. This is a phenomenon where a parasite individual can be passed to a host individual in another way than passing it from parent to offspring. The presence of such host switches can lead to significant incongruences between the separate phylogenetic trees of the hosts and parasites. In our models these separate phylogenetic trees are assumed to be rooted, binary and ultrametric trees, where branch lengths represent the passage of time. The primary objective of this thesis is to reconstruct the joint phylogeny of the hosts and parasites using information coming from their separate phylogenetic trees. To achieve this, we introduce a novel probabilistic tree reconciliation algorithm, which takes the two fully dated trees as input. To the best of our knowledge, it is the only existing tree reconciliation algorithm that works with fully dated trees. This algorithm identifies the horizontal transfers by analyzing differences in the coalescence times and topologies between the host and parasite tree. When a transfer is not fully identifiable, the algorithm makes an informed random choice out of the feasible possibilities, where the different outcomes are weighted by the probability of their occurrence. These probabilities depend on the horizontal transfer rate, denoted by t, which requires an a priori estimate of t based on the separate trees. In this thesis we also provide a novel estimator for this horizontal transfer rate t and analyze some of its mathematical properties such as existence, consistency and asymptotic normality. We evaluate the performance of the algorithm using both simulated and biological datasets, with the help of two distance measures on the space of joint phylogenetic trees. The first is an already existing measure that only considers the topologies of the two joint phylogenetic trees to be compared, while the second is a new distance measure that also accounts for time differences between events, that we propose in this thesis. We also compare the outputs of our reconciliation algorithm with the output of the tree reconciliation software eMPRess. We observe that our reconstruction algorithm performs very well for small rates of horizontal transfer, especially on simulated data. For higher transfer rates the reconstruction becomes more difficult.
Keywords: horizontal transfer; estimation; tree reconciliation; host-parasite model
 
Statistik