invariant sites definition and its effects on tree topology

[Alicia S. Arroyo]

Hello, 

I am studying a gene family using RAxML and using ProtTest in order to know the best model. 

I am wondering what is the real meaning of invariant sites. Does it mean that an invariant site is a position of the alignment that is exactly the same in all sequences? If this is the case but one sequence has a gap in this position, this is still considered an invariant site? 

Let's say we have one invariant site in our alignment, does this position affect to the topology of the tree? I would say no because these invariant sites are not informative enough to make phylogenetic inferences across the taxa. However, we have tested a bit these conditions and we have found that it actually affects both the topology and the bootstrap support. 

Could you please clarify a bit these questions? 

Thank you very much, 

A. 

[Alexandros Stamatakis]

In ML implementations the proportion of invariant sites is estimated via
a free parameter.

Invariant sites do affect the topology since they have the effect to
shrink branch lengths.

For more background on the the topic I'd suggest reading the respective
section in Ziheng Yang's wonderful textbook.

Alexis

On 20.01.2017 14:32, Alicia S. Arroyo wrote:
>
> Hello,
>
> I am studying a gene family using RAxML and using ProtTest in order to
> know the best model.
>
> I am wondering what is the real meaning of invariant sites. Does it mean

> that an invariant site is a position of the alignment that is *exactly*

> the same in all sequences? If this is the case but one sequence has a
> gap in this position, this is still considered an invariant site?
>
> Let's say we have one invariant site in our alignment, does this
> position affect to the topology of the tree? I would say no because
> these invariant sites are not informative enough to make phylogenetic
> inferences across the taxa. However, we have tested a bit these
> conditions and we have found that it actually affects both the topology
> and the bootstrap support.
>
> Could you please clarify a bit these questions?
>
> Thank you very much,
>
>
> A.
>
>
>

> --
> You received this message because you are subscribed to the Google
> Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to raxml+un...@googlegroups.com
> <mailto:raxml+un...@googlegroups.com>.
> For more options, visit https://groups.google.com/d/optout.

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
Adjunct Professor, Dept. of Ecology and Evolutionary Biology, University
of Arizona at Tucson

www.exelixis-lab.org

[edu3716]

Hi Alexis,

Please, could you explain the meaning of a free parameter or at least if the fact that the proportion of invariant sites is estimated via a free parameter have any implication to the topology or to the likelihood of the tree?

I have another question: Can invariant sites affect bootstrap values?

[Alexandros Stamatakis]

On 21.02.2017 11:52, edu3716 wrote:
> Hi Alexis,
>
> Please, could you explain the meaning of a free parameter or at least if
> the fact that the proportion of invariant sites is estimated via a free
> parameter have any implication to the topology or to the likelihood of
> the tree?

both: a different model (with either more or less parameters) might
yield a different tree topology ... otherwise there would be no interest
in dealing with these models at all, the P-Invar model is also known to
affect bracnh lengths

> I have another question: Can invariant sites affect bootstrap values?

of course, since they can affect the tree topology they might affect
branch lengths,

all this information is available in Ziheng's nice textbook I already
mentioned.

alexis

> > an email to raxml+un...@googlegroups.com <javascript:>
> > <mailto:raxml+un...@googlegroups.com <javascript:>>.

> > For more options, visit https://groups.google.com/d/optout

> <https://groups.google.com/d/optout>.

>
> --
> Alexandros (Alexis) Stamatakis
>
> Research Group Leader, Heidelberg Institute for Theoretical Studies
> Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
> Adjunct Professor, Dept. of Ecology and Evolutionary Biology,
> University
> of Arizona at Tucson
>

> www.exelixis-lab.org <http://www.exelixis-lab.org>