Speedemon coalescent rates

[Jacob Christensen]

Greetings,

I'm running speedemon on some SNP data I have and an getting generally good convergence on all parameters. However, on subsequent runs, it appears that a few of the snapperCoalescentRates are switching positions in the log file, which makes me question combining them with LogCombiner. Here you can see it in Tracer:

Does anyone have experience with this?

Best,

Jacob

[Jordan Douglas]

Hi Jacob,

It looks like you have 6 taxa, which means that the first 6 coalescent rates correspond to the leaves, which indices 7-10 are the internal nodes, and 11 is the root. The numbering behind the internal nodes is completely arbitrary between and within runs. Therefore, they are often multimodal in the trace file, which is nothing to be concerned about. The important part is that the leaf and root traces have high ESS (which they generally do). Note that this logic also applies for theta, as well as other models with branch parameters like the relaxed clock.

Jordan

[Jacob Christensen]

Thanks, Jordan. I'm a bit confused about "group count"... I have 6 different species/populations, so should this just be set at 6?

[Jordan Douglas]

Hi Jacob,

I assume you mean 'cluster count'? This is the estimated number of species/subspecies/populations/clusters or whatever they refer to. If all your assigned populations are all split into separate clusters, this term will be 6 in your case. But if they are all lumped together, then cluster count will equal 1.

Jordan

[Jacob Christensen]

No, group count in BEAUti, under the yule skyline collapse model.

[Jordan Douglas]

My bad - the group count is the number of epochs in the Yule skyline model, as detailed below. This parameter is not related to species delimitation 

https://www.beast2.org/2022/06/01/yule-skyline-tree-prior.html