Backbone topology with FBDR model

[crouc...@gmail.com]

Hi all,

I was hoping to get a bit more information on using backbone topologies, specifically with the Fossilized Birth-Death Process for Stratigraphic Range Data model. 

To get familiar with the REvBayes implementation I have been working with the exemplar bears dataset used to demonstrate the FBDR model. In the bears example a single constraint is define (clade_ursinae). Instead of using this constraint I specified a backbone topology of the taxa in that constraint, plus one additional species. I specified this backbone using:

phylogeny <- readTrees("data/backbone.tre")[1]

fbd_tree ~ dnConstrainedTopology(fbd_dist, backbone=phylogeny)

However when executing the model a starting probability cannot be calculated ("Unable to find starting state with computable probability after 1000 tries). I've tried multiple versions of constraint trees with no change. The starting backbone topology used is:

(((((Melursus_ursinus:0.183805722,Ursus_arctos:0.183805722):0.3801097814,Ursus_maritimus:0.5639155033):0.18313593,Helarctos_malayanus:0.7470514333):1.064499815,(((Ursus_americanus:1.228177499,Ursus_thibetanus:1.228177499):0.2652766768,Ursus_abstrusus:1.493454176):0.1026877458,Ursus_spelaeus:1.596141922):0.2154093263):0.5528471427,Arctodus_simus:2.364398391);

I've tried different time scaling of the backbone, different numbers of species etc. Any points would be greatly appreciated.

Nick

[andyf...@gmail.com]

Hi Nick,

Have you tried using fbd_tree.setValue() to provide a starting tree that is compatible with the constraints before running the MCMC?

It is possible that a tree compatible with the given backbone is very unlikely to be drawn at random, and if the problem only manifest when MCMC is run, setValue() will make sure the MCMC starts with a tree that is valid.

Best,

Andy

[crouc...@gmail.com]

Hi Andy,

I have tried adding fbd_tree.setValue(phylogeny) following  fbd_tree ~ dnConstrainedTopology(fbd_dist, backbone=phylogeny), but with the partial backbone tree I get an error message saying that a given tip in the data is not present in the backbone tree. If I try specifying a random topology of all taxa I get the same problem as the original post.  

I would be surprised if the backbone is that unlikely given that it follows the one constraint that is otherwise used in the example analysis provided.

Best,

Nick

[andyf...@gmail.com]

Hi Nick,

Can you share either your full script/inputs or a minimum working example that produces the error you mentioned?

Best,

Andy

[crouc...@gmail.com]

Attached is the example data/code with minimal changes as described above. The mcmc_CEFBDRP_Ranges file is the one that runs the analysis. It currently specifies the backbone_full.tre which is an example backbone of every species. That phylogeny is in the data folder along with the backbone topology specified above.  Specification of the backbone constraint is in the modified model_FBDRP.Rev file

[Jasmine Mah]

Hello! Did you find a solution?