tree collapsing in raxml

[SoniaN]

Hello. I would like to know if there is away within raxml-ng tool in CIPRES to make a majority-rule consensus tree, in a way that would allow me to collapse my best trees by providing a threshold and having all nodes with boostrap values below that threshold be collapsed into polytomies and get a final tree that is annotated with the remaining (above threshold) bootstrap values. Is doing this or anything close to it possible in raxml? Thank you!

[Alexandros Stamatakis]

Dear Sonia,

I don't know about the CIPRES setup as this is not maintained by us.

My guess would be that you will probably have to do this locally on your
computer with RAxML.

In standard RAxML you can easily do what you want to do by specifying an
arbitrary consensus threshold between 51 and 100.

Alexis

On 25.05.21 00:11, SoniaN wrote:
> Hello. I would like to know if there is away within raxml-ng tool in
> CIPRES to make a majority-rule consensus tree, in a way that would allow
> me to collapse my best trees by providing a threshold and having all

> nodes with boostrap values below**that threshold be collapsed into

> polytomies and get a final tree that is annotated with the remaining
> (above threshold) bootstrap values. Is doing this or anything close to
> it possible in raxml? Thank you!
>

> --
> 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>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/raxml/9024fd1f-6274-4b1b-9391-db7c833735d1n%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/9024fd1f-6274-4b1b-9391-db7c833735d1n%40googlegroups.com?utm_medium=email&utm_source=footer>.

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
Affiliated Scientist, Evolutionary Genetics and Paleogenomics (EGP) lab,
Institute of Molecular Biology and Biotechnology, Foundation for
Research and Technology Hellas

www.exelixis-lab.org

[Grimm]

Hi Sonia,

the option Alexi mentioned is -J for classic RAxML (check manual for suboptions), e.g.

raxmlHPC -J T_75 -z trees -m GTRCAT -n XXX

generates a tree with all branches (never "nodes" are supported by ML bootstraps, always internodes!) collapsed respresenting taxon bipartitions found with a frequence less than 75% in the tree-sample, we read in with -z, which in your case would be the bootstrap replicate sample generated by RAxML. If you want a different threshold, just adapt the number (e.g. T_60).

Good consensing, Guido

PS I never got why one would show a consensus tree though as the result of the ML/bootstrap analysis as they are not at all phylogenetic trees but only summary graphs of least-conflicting/unconflicting branching patterns in the pseudoduplicate trees. Their polytomies are, phylogenetically speaking, pretty meaningless. Which I never hesistated to point out to anonymous peer reviewers who wanted me to collapse branches in our trees because of lack of support ;)

PPS If you want to explore the conflict in a tree sample (why BS < 100), consensus networks (which are "cluster union networks", CUNs; a majority rule consensus tree is a 1-dimensional, i.e. reticulation-free, CUN) are more comprehensive summaries of (phylogenetic or BS) tree samples as they show the reason for the decrease: a competing branch in all other replicates or low-frequent, random variations. They can be generated with SplitsTree or the Phangorn library for R using RAxML's BS tree sample as input.

[Alexey Kozlov]

Hi Sonia,

building consensus of a (bootstrap) tree set is implemented in both raxml-ng (--consense option) and
CIPRES portal.

However, if you want to collapse low-support nodes in the *best ML tree*, this is not directly
implemented in raxml-ng/CIPRES. But it is a really trivial operation, which can be done even in a
tree viewer (eg in Dendroscope: Edit -> Contract low support edges).

Best,
Alexey

On 25.05.21 14:53, Grimm wrote:
> Hi Sonia,
>
> the option Alexi mentioned is -J for classic RAxML (check manual for suboptions), e.g.
>
> raxmlHPC -J T_75 -z trees -m GTRCAT -n XXX
>
> generates a tree with all branches (never "nodes" are supported by ML bootstraps, always
> internodes!) collapsed respresenting taxon bipartitions found with a frequence less than 75% in the
> tree-sample, we read in with -z, which in your case would be the bootstrap replicate sample
> generated by RAxML. If you want a different threshold, just adapt the number (e.g. T_60).
>
> Good consensing, Guido
>
> PS I never got why one would show a consensus tree though as the result of the ML/bootstrap analysis
> as they are not at all phylogenetic trees but only summary graphs of least-conflicting/unconflicting

> branching patterns in the *pseudo*duplicate trees. Their polytomies are, phylogenetically speaking,

> pretty meaningless. Which I never hesistated to point out to anonymous peer reviewers who wanted me
> to collapse branches in our trees because of lack of support ;)
>
> PPS If you want to explore the conflict in a tree sample (why BS < 100), consensus networks (which
> are "cluster union networks

> <https://phylonetworks.blogspot.com/2020/04/consensus-networks-cluster-union-or_6.html>", CUNs; a

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

>
> --
> 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>.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/raxml/427e4fb1-64ed-47b0-9658-6595cb6b6974n%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/427e4fb1-64ed-47b0-9658-6595cb6b6974n%40googlegroups.com?utm_medium=email&utm_source=footer>.