SH like support

[Karen]

Dear all,

I found in a post that the SH-like support (-f J) refers to  Anisimova & Gascuel 2006 (http://sysbio.oxfordjournals.org/content/55/4/539)

correct!?  I am wondering whether in RAxML I also can perform the approximate likelihood ratio support (aLRT) published later in

PHYML  (Guindon et al., 2010, http://sysbio.oxfordjournals.org/content/59/3/307) ? (what is the exact difference!?)

(With IQtree, where I can perform both)

Thanks Karen

[Alexandros Stamatakis]

Hi Karen,

Please see this paper here:

http://sysbio.oxfordjournals.org/content/60/5/685

that surveys all of these methods.

RAxML actually only implements the SH-like aLRT that was published later
I think, I must have provided the incorrect reference in that post.

Alexis

On 17.01.2016 16:41, Karen wrote:
> Dear all,
>
>
> I found in a post that the SH-like support (-f J) refers to Anisimova &
> Gascuel 2006 (http://sysbio.oxfordjournals.org/content/55/4/539

> <http://sysbio.oxfordjournals.org/content/55/4/539>)

>
> correct!? I am wondering whether in RAxML I also can perform the

> approximate likelihood ratio support (SH-aLRT) published later in
>
> PHYML (Guindon et al., 2010 <http://dx.doi.org/10.1093/sysbio/syq010>,
> http://sysbio.oxfordjournals.org/content/59/3/307
> <http://sysbio.oxfordjournals.org/content/59/3/307>) ? (what is the

> exact difference!?)
>
> (With IQtree, where I can perform both)
>
>
> Thanks Karen
>

> --
> 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

[Heiko]

Hi Karen,

the main difference between the aLRT and the SH-aLRT is how the Null distribution of likelihood ratios tested against are determined.

In aLRT a 'standard' likelihood ratio rest (LRT) is used to figure out which of the 3 possible resolutions on a branch is most supported. However, the 3 tree topologies are not nested - one of the major requirements to be allowed to use the Chi2 distribution as an approximation of the Null distribution. Hence, the test is performed against the star tree at this branch, because that one is nested in (i.e. is a special case of) all three resolved quartets topologies at that branch.

In SH-aLRT they overcome the problem of non-nestedness by applying the machinery of the Shimodairo-Hasegawa test (SH) to get the Null distribution. This allows them to test non-nested topologies directly.

Cheers,
Heiko

[Karen]

Many thanks Heiko! :)
Cheers Karen

[Karen]

Many thanks Alexis!

Thanks for clarification. Makes sense because in the Manual you refer to
http://sysbio.oxfordjournals.org/content/59/3/307.long
(which is the similar thing implemented in IQtree as aLRT -1000)
I was just confused about the different references in post and manual.

Cheers Karen