Topology constraints possible?

[Ziad Khouri]

Hi,

I wanted to ask whether there is currently a way to set partial topology constraints (e.g. enforce the monophyly of a specific group without fixing the entire tree topology).

Thank you!

Best,

Ziad

[Benjamin Redelings]

Hi Ziad,

Hmm... not right now.

Would you like me to put this back in?  I guess the way it would work is that you would specify a multifurcating tree, and any internal branch imposes a constraint.

Are you thinking that it would make the program faster?

-BenRI

--
You received this message because you are subscribed to the Google Groups "bali-phy-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bali-phy-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bali-phy-users/0c741c97-3088-4cfb-ae72-eb88c631c686n%40googlegroups.com.

[Benjamin Redelings]

Hi Ziad,

I thought of a way to implement this that is relatively CPU efficient.  Basically, each constraint can correspond to a virtual binary character, and we can then constrain the parsimony score to be  <= 1 for each virtual character.

What do you think?

-BenRI

[Ziad Khouri]

Hi Ben,

Thank you for getting back to me and apologies for the late response. The use case I was thinking of is something like this:

There is 1 relatively short locus of interest for which we want to estimate alignment + phylogeny (or estimate a parameter of a codon model, or reconstruct ancestral states, etc.). However, using this locus alone means high uncertainty with respect to tree topology. One could add many other loci to the analysis, but at some point, this starts taking too long to run. If certain relationships (i.e. specific splits) are well established based on previous studies with large datasets, the ability to constrain those relationships in the current analysis could be an efficient way to incorporate external information without too much of a performance hit and hopefully improve the precision with which parameters of interest are estimated (of course with the caveat that everything is now conditioned on a "fact" that might not be true).

I am unfortunately not familiar with the inner workings of BAli-Phy, so not sure about what would be an efficient implementation. In terms of functionality from a user perspective, I think it would be ideal to be able to set "partial" or "backbone" constraints analogous to those possible in MrBayes. For a split of interest, specify three sets of taxa: one set that goes on one side of the split, another set that goes on the other, and the remaining taxa that can go on either "side". Using the virtual character approach you suggested, would this work by giving one set of taxa state 0, another set state 1, and the remaining coded as "?"?

Best,

Ziad