Skyline Plots

Bayesian Analysis requires to put priors on everything

What are tree priors? 

• Backwards in time Coalescent processes
• Forward in time Birth-Death processes

Tree priors are tree generating processes

Changing the number of bubbles through time can be used to model population dynamics

Drummond et al. 2003

Birth-death models model the birth, death and sampling of lineages

Coalescent

• Conditions on sampling, assumes only a small portion of the population is sampled
• Effective population size corresponds to "size of a population"
• 1 parameter to infer

Birth-Death

• models the sampling process
• birth and death rates correspond to how "a population is changing" (growing declining)
• 3 parameters out of which two can be inferred.

How do birth-death and coalescent models differ?

• In the way the times of rate changes are defined.
• In the smoothing priors.

How do birth-death and coalescent models differ

Skyline plots allow to have different rates or effective population sizes at different points in time.

• In the way the times of rate changes are defined.
• In the smoothing priors.

How do different skyline models differ?

https://taming-the-beast.org/tutorials/Skyline-plots/

Tutorial on the Bayesian skyline plot and the birth death skyline model