link tree model and unlink clock models

[Morgan Geniviva]

Hi all,

I am running a BEAST analysis on HIV-1 env sequences with partitions to account for recombination. At first I had the substitution models, clock models, and tree models linked but I want to try unlinking the tree models to account for recombination. When I clicked on unlink tree models in BEAUTi it automatically unlinked the clock models as well and then when I clicked link clock models it automatically linked the tree models. I was able to manually link the clock models by renaming the model partitions.

So my question: is this a bug in BEAUTi or is there a reason why I shouldn't unlink the tree models and link the clock models?

[Lambodhar Damodaran]

Hi Morgan,

I believe it has to do with the fact that unlinking the tree model essentially treats each partition as its own tree building process (making its own tree). Because you are partitioning the the sites but want to account for recombination you want to make sure that your estimating for one tree topology. Also as a side note have you already detected recombination/ know there are recombinant sequnces in your dataset? It would probably be good to run RDP4 for good measure to determine if you really need to do multiple partitions.

Best,

Lambo

[Morgan Geniviva]

Hi Lambo,

Thank you for that clarification! Yes, I have already detected recombination in my dataset. My initial plan was to run my sequences through RAPR and remove any recombinant sequences however this left me with 6 sequences (versus the 382 that I started with) so I ran the sequences through GARD to identify potential breakpoints. 

Best,

Morgan

[Pavel Rinkman]

Hello, Morgan Geniviva,

My name is Pavel, and I have been delving into HIV phylogenies using BEAST. I initially tried contacting you directly via email but faced some delivery issues.

HIV is a new object for me. So I am keen to understand more about your research. Are you investigating a local outbreak or the global history of a specific HIV lineage? Does your study touch upon phylogeography?

We might encounter the same problems during the dataset preparation and BEAST analysis setting. Maybe it would be interesting for you to discuss these points and found approaches to solve these problems?

For instance,  the nuances of selecting suitable prior parameters and models. Which publications do you consider trustworthy for choosing specific parameter values? Which recombination detection tool did you finally use?

I am working on the geographical distribution history of HIV in Armenia. I would be open to sharing my code and experience for solving some tasks if that could assist anyone.

Thank you all for your time.

Best regards,

Pavel

P.S. I am sorry for the off-topic.

вторник, 4 июля 2023 г. в 23:03:39 UTC+3, Morgan Geniviva:

[Andrew Rambaut]

Sorry for the slow response - it has already been mentioned in a reply that you can’t have linked clock models on unlinked trees because the clock models are specific to a tree. 

However, you can link the parameters of the clock models in the priors table if you want the partitions to have the same rate (or the same stdev). Another possibility if you have lots of partitions with different clock rates is to collect them into a hierarchical model (also done in the priors table).

https://beast.community/hierarchical_models

Best,

Andrew

--
You received this message because you are subscribed to the Google Groups "beast-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/beast-users/6d7f4aa5-0fa7-44e0-9372-1e8a7ec6405dn%40googlegroups.com.