Backbone topology, non-FBD

[Zachary Griebenow]

Greetings all, I am looking into incorporating a backbone argument in a non-FBD total-evidence analysis. However, dnConstrainedTopology, which is the only context in which I am familiar with backbone, also requires an FBD distribution as an argument. Is it possible to implement a backbone topology whatsoever outside the context of an FBD analysis (that is, without using dnConstrainedTopology), or would I need to resort to clade constraints to accomplish the same end?

Regards, Zach

[David Černý]

Hi Zach,

It's not true that dnConstrainedTopology requires FBD as the base distribution; it's actually supposed to be very general. For unrooted (nonclock) trees, it's supposed to work with any base distribution that can be cast to (= is derived from) UniformTopologyBranchLengthDistribution – and it sort of does, except that you might have to provide your own initial tree (which itself has to satisfy the backbone constraint, of course) to get the analysis started. See issue #384 in the GitHub repo. For rooted (clock) trees, it will work with any base distribution that can be cast to AbstractRootedTreeDistribution, which would be true of most flavors of the birth-death process. Even for a distribution like dnUniformTimeTree, which is not derived from AbstractRootedTreeDistribution, it's still possible to make it work if you supply your own starting tree. See issue #512 in the GitHub repo. I think the only base distributions that are currently known not to work with dnConstrainedTopology are dnCDBDP and dnGLHBDSP – see issue #375 in the GitHub repo.

Hope this helps!

El miércoles, 14 de agosto de 2024 a la(s) 3:42:24 p.m. UTC-5, zachary....@colostate.edu escribió:

Greetings all, I am looking into incorporating a backbone argument in a non-FBD total-evidence analysis. However, dnConstrainedTopology, which is the only context in which I am familiar with backbone, also requires an FBD distribution as an argument. Is it possible to implement a backbone topology whatsoever outside the context of an FBD analysis (that is, without using dnConstrainedTopology), or would I need to resort to clade constraints to accomplish the same end?

Regards, Zach

Best,

--

David Černý