Beast questionssss

[Celine]

Dear all,

I am a new Beast user and I have some troubles.
My aim is to obtain an ultrametric tree and use it in apes for
diversification rate estimations. I work on two different genes (COI
and a “functional” gene belonging to a multigene family) extracted
from the same specimens. I would like to compare the diversification
rate of this two genes in the same range of time (even if this time is
not important for me).
I have no idea about rate of evolution of these two genes and I have
no precise calibration point (but my aim is not to have absolute dates
but just relatives “dates”). I obtained trees from MrBayes and Ml
analyses and there are congruents for the two genes. I would like to
use these topologies in beast.
According to this study and the Beast utilisation and more precisely
Beauti parameters I have few questions:
- In the model panel: If I would like to partition into codon position
((1st+2nd) and (3rd)), chose the SRD06 model which model parameters
entered will be used for the partitioned data. How can I assign
different parameters to the partitioned data?
- I said above that I have not a precise calibration but I can have a
very approximate idea of the root. Given I am not interested in
absolute date but instead in relative dates, Will the point
calibration have some consequences in the relative dates? I think not
but it may be wrong.
If it is a problem and I fix the mean substitution rate to 1, can I
have an ultrametric tree?
- I would like to use a starting tree and fixed the topology during
all the analysis. I think I have to insert in the xml file <newick
id="startingTree"> (tree in newick format); </newick> instead of
<!-- Generate a random starting tree under the coalescent
process -->
<coalescentTree id="startingTree"> <constrainedTaxa> <taxa
idref="taxa"/> <tmrca monophyletic="true"> <taxa idref="ingroup"/> </
tmrca> </constrainedTaxa> <constantSize idref="initialDemo"/> </
coalescentTree>.
And to fix the tree topology I have to disactivate in operators panels
all tree rearrangements. and Is it correct?
In the starting trees, branch lengths are considered in substitution/
site, how these branch length are used by BEAST?

I am sorry for all these questions and I hope someone will answer me
(rapidly…).
Many thanks in advance.

Céline

[Simon Ho]

Hi Céline,

Here are some answers to your questions:

> -       In the model panel: If I would like to partition into codon position
> ((1st+2nd) and (3rd)), chose the SRD06 model which model parameters
> entered will be used for the partitioned data. How can I assign
> different parameters to the partitioned data?

Do you mean you want to use a different substitution model for the two
data partitions (for example, HKY for 1st+2nd and GTR for 3rd)? You
will need to make a few edits to the XML file. Perhaps one way to
start is to generate one input file with HKY for both partitions, then
another input file with GTR for both partitions, then copy and paste
some of the elements. You need to make changes to the following
elements: "substitutionModel", "siteModel", "operators", and the
parameters that are logged to the output file. Let me know if you run
into any problems and I can send you an example.

> -       I said above that I have not a precise calibration but I can have a
> very approximate idea of the root. Given I am not interested in
> absolute date but instead in relative dates, Will the point
> calibration have some consequences in the relative dates? I think not
> but it may be wrong.
> If it is a problem and I fix the mean substitution rate to 1, can I
> have an ultrametric tree?

It is better to fix the substitution rate to 1. BEAST will produce
ultrametric trees.
An alternative is to put a prior on the root (e.g., normally-
distributed prior), but enforcing a point calibration is probably not
a good idea.

> -       I would like to use a starting tree and fixed the topology during
> all the analysis. I think I have to insert in the xml file <newick
> id="startingTree"> (tree in newick format);        </newick> instead of
>  <!-- Generate a random starting tree under the coalescent
> process            -->
>         <coalescentTree id="startingTree">      <constrainedTaxa> <taxa
> idref="taxa"/>     <tmrca monophyletic="true">     <taxa idref="ingroup"/>         </
> tmrca>       </constrainedTaxa>        <constantSize idref="initialDemo"/>     </
> coalescentTree>.

Yes, that's correct.

> And to fix the tree topology I have to disactivate in operators panels
> all tree rearrangements. and Is it correct?

Also correct. By default, there are four tree topology operators
(subtreeSlide, narrowExchange, wideExchange, and wilsonBalding). These
all need to be removed if you want the tree topology to be fixed.

> In the starting trees, branch lengths are considered in substitution/
> site, how these branch length are used by BEAST?

The interpretation of the branch lengths depends on the units used in
your analysis. If you are fixing the rate to 1, then providing a
starting tree with branch lengths in subs/site should not be a
problem.

Cheers,
Simon

[Celine]

Hi Simon and beast users,

First many thanks for your answers.
Despite your answers were really precise, I have a problem with codon
partition. I generated the two xml files, one with substitution model
and parameters for (1st+2nd) codon positions and one for the 3rd codon
positions. Now how can I specifically attributed each model for each
partition?
I edit xml file like that:
<!-- The HKY substitution model (Hasegawa, Kishino & Yano,
1985) -->
<hkyModel id="hky1">
<frequencies>
<frequencyModel dataType="nucleotide">
<alignment idref="alignment"/>
<frequencies>
<parameter id="hky1.frequencies" dimension="4"/>
</frequencies>
</frequencyModel>
</frequencies>
<kappa>
<parameter id="hky1.kappa" value="1.0" lower="1.0E-8"
upper="Infinity"/>
</kappa>
</hkyModel>
<!-- The HKY substitution model (Hasegawa, Kishino & Yano,
1985) -->
<!-- The general time reversible (GTR) substitution
model -->
<gtrModel id="gtr2">
<frequencies>
<frequencyModel dataType="nucleotide">
<alignment idref="alignment"/>
<frequencies>
<parameter id="gtr2.frequencies" dimension="4"/>
</frequencies>
</frequencyModel>
</frequencies>
<rateAC>
<parameter id="gtr2.ac" value="1.0" lower="1.0E-8" upper="Infinity"/
>
</rateAC>
<rateAG>
<parameter id="gtr2.ag" value="1.0" lower="1.0E-8" upper="Infinity"/
>
</rateAG>
<rateAT>
<parameter id="gtr2.at" value="1.0" lower="1.0E-8" upper="Infinity"/
>
</rateAT>
<rateCG>
<parameter id="gtr2.cg" value="1.0" lower="1.0E-8" upper="Infinity"/
>
</rateCG>
<rateGT>
<parameter id="gtr2.gt" value="1.0" lower="1.0E-8" upper="Infinity"/
>
</rateGT>
</gtrModel>

<!-- site
model -->
<siteModel id="siteModel1">
<substitutionModel>
<hkyModel idref="hky1"/>
</substitutionModel>
<relativeRate>
<parameter id="siteModel1.mu" value="1.0" lower="0.0"
upper="Infinity"/>
</relativeRate>
<gammaShape gammaCategories="4">
<parameter id="siteModel1.alpha" value="0.5" lower="0.0"
upper="Infinity"/>
</gammaShape>
</siteModel>
<!-- site
model -->
<siteModel id="siteModel2">
<substitutionModel>
<gtrModel idref="gtr2"/>
</substitutionModel>
<relativeRate>
<parameter id="siteModel2.mu" value="1.0" lower="0.0"
upper="Infinity"/>
</relativeRate>
<gammaShape gammaCategories="6">
<parameter id="siteModel2.alpha" value="0.5" lower="0.0"
upper="Infinity"/>
</gammaShape>
</siteModel>

<compoundParameter id="allMus">
<parameter idref="siteModel1.mu"/>
<parameter idref="siteModel2.mu"/>
</compoundParameter>

Do you think that it is correct?

Thank you again!

Regards,

Céline

[Simon Ho]

Hi Céline,

That looks OK to me.
Don't forget to check that you have operators for the various model
parameters.

Cheers,
Simon