What is difference between linked and unlinked partitions?

[Anne]

Hi everyone,

I didn't understand what the difference between linked and unlinked partitions. When must I use linked? For example, if genes have been evolved under same model can I link them? And trees, when I link the trees what is used to build the final tree? Would be used tmrca?

Thank you!

[Andrew Rambaut]

By default, partitions start as linked substitution models, clock models and trees. The only difference between this and just
concatenating the sequences is that they will have relative rate parameters ('mu' parameters in the log files). By partitioning
you are saying they have different rates of evolution but share everything else.

If you unlink substitution models you are saying they have different substitution models (you can specify what these are independently
for each partition) and independent parameters of those substitution models.

If you unlink clock models then you are saying the partitions have different distributions of rates amongst branches (i.e., some branches
can be faster for some genes and slower for others).

If you unlink trees then you are saying that each partition evolved with a completely independent tree. This might be the case if the
partitions are unlinked (in the recombination sense) loci.

Hope this helps,
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 post to this group, send email to beast...@googlegroups.com.
> Visit this group at http://groups.google.com/group/beast-users.
> For more options, visit https://groups.google.com/groups/opt_out.

___________________________________________________________________
Andrew Rambaut
Institute of Evolutionary Biology University of Edinburgh
Ashworth Laboratories Edinburgh EH9 3JT
EMAIL - a.ra...@ed.ac.uk TEL - +44 131 6508624


--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

[Natalia]

Hello,

I have a similar question with Anne. I have 6 different genes and I want one tree with one set of branchlengths common for all 6 genes. Does this mean that I need to make one clock or unlink clocks? I want to use an uncorrelated lognormal clock. Also, I have an estimate for the lognormal distribution of the rate of one gene and I would like to put it as a prior. Wouldn't this necessitate to unlink the clocks? and if I unlink them and click estimate for all the other rates, will they be "scaled" to the one I have the prior for? (i.e. will there be one set of branchlenghts?). 

I have also a minor question. Two of my partitions have almost identical GTR rate estimates and I would like to merge them to one so I have less parameters. However, they have very different Gamma estimates. Is there a way to preserve the different Gamma parameters for them, but have the rest of the GTR model common?

thank you,

Natalia