Possible addition of "BOOSTER" - Bootstrap support by Transfer (Lemoine et al, Nature, 2018)?

[Frank Wright (BioSS)]

Hi,

I came across the BOOSTER method that seeks to modify Joe Felsenstein's bootstrap method for the era of Big Data.  This was after comments on this list about the performance of the bootstrap on long phylogenomic datasets - these stimulated my curiousity.  I then came across BOOSTER which is looking instead at "moderate signal".

QUOTE "With large trees and moderate phylogenetic signal, BOOSTER tends to be more informative than the standard Felsenstein bootstrap method".  It is discussed in an April 2018 paper by Lemoine et al:

https://www.nature.com/articles/s41586-018-0043-0

There is also a BOOSTER webpage:

https://booster.pasteur.fr/

which mentions that the method is implemented in PhyML and RAxML-NG and also provides a link to the BOOSTER source code on Github.

Perhaps it is already on the things-to-add list IQ-TREE?  I guess it is already a very long list but it would be great if BOOSTER could be added.

Thanks again for proding such a brillant piece of software,

Frank Wright

BioSS Edinburgh

[Heiko Schmidt]

Dear Frank,

Thank you for your interest and suggestion.

If I understood Minh correctly, the transfer bootstrap expectation (TBE) values are currently tested in the latest beta version of IQ-Tree (available from the IQ-Tree Github repository).

Best wishes,
Heiko

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

-----------------------------------------------------------------------------
Heiko Schmidt
Center for Integrative Bioinformatics Vienna (CIBIV)
University of Vienna / Max F. Perutz Laboratories (MFPL)
http://www.cibiv.at/
-----------------------------------------------------------------------------

[Minh Bui]

Yes, Heiko is right. The beta version has a new option --tbe to perform TBE after standard bootstrap. You can download from here: https://github.com/Cibiv/IQ-TREE/releases/tag/v1.7-beta. For example you can do

iqtree -s ALN_FILE -b 100 --tbe

will create 2 more files: ALN_FILE.tbe.tree (for ML tree with TBE supports) and ALN_FILE.tbe.stat (for some statistics to identify rogue taxa).

Cheers

Minh

[Antony Vincent]

Hi,

TBE cannot works with UFBoot (-bb) ?

Antony

[Minh Bui]

Hi Antony,

That’s right. Because TBE was designed to correct Felsenstein’s bootstrap, not UFBoot. In fact, UFBoot already has a bias correction, thus doing a double correction (UFBoot+TBE) is not valid. I clarified this with Booster author (https://github.com/evolbioinfo/booster/issues/1). 

Cheers

Minh

[Antony Vincent]

Thank you for your answer !

[Mark Miller]

Dear Minh,

We are interested in getting the TBE feature released in CIPRES, but we usually do not release beta versions into production.
Do you know what the timeline is for going form beta to full release for 1.7? And do you think there are any concerns for putting it into production to a broad audience?

Best,

Mark

[Minh Bui]

Dear Mark,

On 10 Nov 2018, at 4:22 am, Mark Miller <mark.ala...@gmail.com> wrote:

Dear Minh,

We are interested in getting the TBE feature released in CIPRES, but we usually do not release beta versions into production.
Do you know what the timeline is for going form beta to full release for 1.7?

Thanks for asking. Ver 1.7 release plan is February next year. There are a couple of features we want to put in for this release.

And do you think there are any concerns for putting it into production to a broad audience?

More testings are still needed...

Cheers,

Minh

[Vigneshwaran Renganathan]

Dear Minh,

How to use  --tbe option with pre generated .bootstraptrees files using  "iqtree  -s input-alignmentfile  -bo 20 -pre boot1"

is it available to use with "iqtree -con -t alltrees"?

Thanks

Vignesh

[Minh Bui]

Hi Vignesh,

This not is possible in IQ-Tree yet, but can you use the Booster code directly for this: 

https://github.com/evolbioinfo/booster

Minh

[jp.fland...@gmail.com]

Hi Minh,

I would appreciate some clarification concerning this point. 

TBE is computed from a set of bootstrap trees and in the cas of UFB we have also a set of 'FSB-like'-bootstrap trees. This set is conceptually similar to the set of Felsestein bootstrap trees.  After a careful reading of the first and second paper concerning UFB, I cannot find the indication of a 'correction' of the set of trees or individual UFB tree. Even if there are build under another assumptions and by another way, the UFB bootstrap trees set remains a set of "possibles" trees. So I do not understand why TBE cannot be computed from the UFB bootstrap set. I think that the early position of the BOOSTER team was close to mine and that the answer you received was mainly due to a misunderstanding. Of course I agree that much precautions have to be taken in the interpretation as the demonstration of the interest of TBE is dealing only on FSB.

JPF

[mbfia...@gmail.com]

Hi, I also had the same thoughts and referenced back to this question on an issue at the github repository. The only correction I could find on the UFBoot paper was concerning polytomies, but I couldn't say if this correction would be enough to invalidade using TBE with UFBoot

Matt

[Minh Bui]

Hi Matt,

There was a follow-up discussion here (in next thread):

https://groups.google.com/forum/#!topic/iqtree/HAxyKySfAXI

So it is not just about polytomies. As JPF termed it “quasi-bootstrap trees”, which can’t be used as input for TBE, which work on Felsenstein’s bootstrap trees.

Cheers,

Minh

[mbfia...@gmail.com]

Hi again Minh,

I've re-read (once more) the UFBoot paper and the discussion you just linked and I think I have a clearer image of the problem of UFBoot with TBE and why it's a bad idea.

Thanks you for your patience enlightening me!

Regards,

Matt 

Minh

To unsubscribe from this group and stop receiving emails from it, send an email to iqt...@googlegroups.com.