Rerooting on ancestral sequence before computing supports
However, ....
BEAST support inference is still not performing well. We spoke with Jiansi, and it sounds like the issue is stemming from the fact that our inference model is not aligned with our simulation model. Specifically, BEAST estimates time trees, whereas we are simulating topology trees and inferring support for topological nodes. I suggest we table the BEAST comparison for now and consider other methods that have more easy options for specifying time trees like MrBayes.
Includes:
.treefile.However, ....
BEAST support inference is still not performing well. We spoke with Jiansi, and it sounds like the issue is stemming from the fact that our inference model is not aligned with our simulation model. Specifically, BEAST estimates time trees, whereas we are simulating topology trees and inferring support for topological nodes. I suggest we table the BEAST comparison for now and consider other methods that have more easy options for specifying time trees like MrBayes.