Paralogous partitions?

[Aaron Lewis]

Hello,

I see that BAli-Phy supports model parameter sharing between partitions, but it's not clear to me if it would also support sharing inferred sequences between partitions. I'm working with a gene that has undergone duplication, and would like to process it as a pair of partitions that share a common node, to take full advantage of prior knowledge that they share both the same tree topology and a common ancestor. This would be preferable to simpler but less powerful alternatives, like including all sequences in a single partition and thereby losing the tree constraint, or imposing alignment to a fixed ancestral sequence that could be shared by both partitions, and thereby losing the ability to marginalise over alignments. Is this an option?

Cheers

Aaron Lewis

[Benjamin Redelings]

Hi Aaron,

Have you looked at papers that attempt to reconcile gene trees to species trees by adding duplication and loss events?  It sounds like you would like to use a duplication/loss model and assume exactly 1 duplication and 0 losses.  Are you completely sure that there are no other duplications or losses?  If the tree for the two groups is not the same, one interpretation is that in fact there are additional duplications and losses.

Unfortunately, BAli-Phy doesn't have duplication-loss models yet.  And you can't force the subtrees to be equal by using partitions.  Usually this is not necessary though, unless perhaps the genes are very short.

However, it seems to me like you should include all the sequences in the same partition.  If the posterior probability that both groups have the same tree is high enough, then perhaps that the 1-duplication + 0-losses model is OK.  In that case, then you can condition on the two sub-trees matching by the simple procedure of selecting only the posterior samples that satisfy that condition.  (This is an allowed way of constructing a conditional posterior.)  I guess you would need to write a python script or something to identify which samples satisfy the condition though.

Does that help?

-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/CAD5NXAL_oeGJ-4emARo%2BfeFEm%2B%3DcxKxvv8Ur1xX-RfqGjJB4Lw%40mail.gmail.com.

[Aaron Lewis]

Hi, Ben, and thanks for the tips, particularly the suggestion to condition on the subtrees matching. That could be a good post-hoc way to get just what I was looking for. Cheers,

Aaron