Seminar - Estimating dated phylogenetic trees with applications in epidemiology and macroevolution. Machine learning methods for genotype-phenotype association studies based on non-coding DNA variation

School of Mathematics and Statistics Research Seminar

Speaker: Dr Alexandra Gavryushkina
Time: Wednesday 19th February 2020 at 04:00 PM - 05:00 PM
Location: Cotton Club, Cotton 350
Groups: "Mathematics" "Statistics and Operations Research"

Add to Calendar Add to your calendar

Abstract

Newly available biological data require developing new approaches to reconstructing dated phylogenetic trees. In the first part of the talk, I will present new methods that employ birth-death-sampling models to reconstruct dated phylogenetic trees in a Bayesian framework. These methods have been successfully applied in epidemiology and macroevolution. Dated phylogenetic histories can be informative about past events, for example in epidemiology, we can learn from a reconstructed transmission tree which individuals were likely to infect other individuals. By reconstructing dated phylogenetic trees, we can also learn about the tree generating process parameters. For example, we can estimate and predict how fast epidemics spread or how fast new species arise or go extinct.

Mashing learning methods have been used to identifying mutations in protein coding regions of the genomes that are associated with phenotypic changes. However, in most species the major part of the genome consists of regions that do not code for proteins. These non-coding regions often play important roles in different functions of organisms and therefore mutations in these regions can lead to phenotypic changes. We study whether these methods can be extended to non-coding DNA variation.

Go backGo back to the seminar list