Reversible jump model selection option in BEAST2

[Jacob Berv]

Hi there

Does anyone know if the RB model selection option in beast2 will auto partition the data? In Bouckaret et al 2012, this auto-partition procedure is described briefly:

"We propose a new model where we partition the genome in k consecutive parts. The boundaries of the parti- tions is randomly assigned and changed throughout the MCMC run. For each partition, a different substitution model is used. Since we do not know which substitution model is appropriate beforehand, we estimate the sub- stitution model using a reversible-jump based method. The substitution model jumps between the following models: Felsenstein [23] (F81), HKY, Tamura and Nei [24] (TN93), transitional model (TIM), symmetrical model (SYM) and generalised time-reversible (GTR). The number of free parameters increases from 0 for F81 to 5 for GTR, and these parameters have a gamma (0.2, 5) prior each." 

But it's not clear if this is implemented in the rb-beast package in the StarBeast mode. 

Also - does anyone know if the RB model selection procedure searches through and of the +I +G model space? I have the option of telling BEAST to estimate the gamma shape and the proportion of invariant AND use the RB model - but it's not clear what this will do. Any thoughts?

Jake

[Remco Bouckaert]

Hi Jake,

The auto-partition model is in the RBS package. There is a BEAUti template for doing a standard analysis, where the partition is automatically split into a given number of partitions. You can use it in BEAUti by choosing the File/Template/AutoPartition template. To use it in combination with a *BEAST analysis requires some XML editing. Note that there is an issue at the moment with resuming an AutoPartion analysis at the moment.

The RBS package does not provide a way to search through gamma rate heterogeneity or invariant sites.

Cheers,

Remco