Re: [raxml] Bootstrap values

2,508 views
Skip to first unread message
Message has been deleted

William Gearty

unread,
Oct 6, 2014, 12:58:50 PM10/6/14
to ra...@googlegroups.com
You could try making a consensus tree with the bootstraps.
It's not going to change the result, but it will collapse any nodes that have low support values.
In terms of actually getting higher bootstrap values, I'm not sure.
If you are running 1,000 bootstraps and still getting low support, I would suggest that that is the signal of your alignment, and you can't really change that.
This is a molecular alignment?
Nuclear, mitochondrial? Coding, non-coding?

On Mon, Oct 6, 2014 at 9:54 AM, Nick <anikets...@gmail.com> wrote:
Hi,

I ran RAxML-HPC2 at XSEDE (05 Oct 2014) on the Cipres portal. The only thing that I had changed from the default settings was the number of bootstrap iterations (increasing to 1,000).

I had a few questions. I am attaching the relevant files in case you needed to have a look.

1. I am trying to get figtree display the bootstrap values. If I Iook into the node labels of the "RAxML_bipartition" file, that should give me the bootstrap values, right?
2. I tried this and the values look pretty low, as low as 1, but sometimes as high as 100. Other than the possibility that my alignment has conflicting signals, is there any thing in the run itself that I can change, like a parameter, etc., to get a more confident tree?

Best,
Nick

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



--
Message has been deleted

William Gearty

unread,
Oct 6, 2014, 4:14:19 PM10/6/14
to ra...@googlegroups.com
Hi Nick,
So the way that RAxML makes that bipartitions tree is that it takes the best tree from the ML search and then places the bootstraps for each node (calculated from the bootstrap analysis) on that tree.
Often if there are lots of topologies that have similar likelihoods, you might get a best tree that has nodes that don't often actually pop up over the course of the bootstrap analysis.
I am guessing something like this is happening, which is giving you low support values.
Therefore, I would suggest making a consensus tree of your bootstrap trees (I'm not sure what the commands are, but try looking in the manual).
I'm not aware of anything about plant DNA being super conflicting, but perhaps you are aware of something?
Hope this helps,
Will

On Mon, Oct 6, 2014 at 12:53 PM, Aniket Sengupta <anikets...@gmail.com> wrote:
Hi,

About my first question -- so, I am doing it right, correct? Those are the bootstrap values.
About the alignment: it is a plant nuclear protein coding sequence (DNA).
I understand that the alignment might be generating conflicting signals but 1/100 support is a little uneasy in any analysis. If it were at least a few more, it would have made sense, but here it looks like only one out of hundred trees support that node.

Does that help?

Best,
Nick

--
You received this message because you are subscribed to a topic in the Google Groups "raxml" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/raxml/ir9c3vYVIWs/unsubscribe.
To unsubscribe from this group and all its topics, send an email to raxml+un...@googlegroups.com.

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



--
Aniket Sengupta,
Doctoral Aspirant,

The Hileman Lab of Plant Evolution and Development,

The Department of Ecology and Evolutionary Biology,

The University of Kansas,

Haworth Hall, Room 8022,

1200 Sunnyside Avenue,

Lawrence, Kansas 66045.

anikets...@gmail.com
----------------------------------------------
<^v^>   What immortal hand or eye / Could frame thy fearful symmetry? -William Blake

--
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.
For more options, visit https://groups.google.com/d/optout.
Message has been deleted

William Gearty

unread,
Oct 6, 2014, 6:44:50 PM10/6/14
to ra...@googlegroups.com
I believe those are the relative node ages. I don't know too much about how RAxML calculates those, but I assume that those values are just summations of the branch lengths.
You can check exactly what is being shown in FigTree by expanding the different options on the left (such as Node Labels or Branch Labels).
I have definitely seen "best trees" shown in publication, although I only see them published with support values if those support values are relatively high. Generally, for maximum likelihood, I think people only really care about 80 or 90+ support values. Although that varies from person to person.
-Will

On Mon, Oct 6, 2014 at 2:53 PM, Aniket Sengupta <anikets...@gmail.com> wrote:
Hi,

I will try that. One more -- when I import the bi-partition  files into figtree, other than branch-lengths,  some decimal values pop up at the nodes. They are sometimes greater than one. They can not be BS support as BS is a proportion that can't exceed 1. Do you know what these values are?

A more basic question. Do people publish the "best tree" aka, the tree that maximizes the likelihood even if the BS is as low as 1? Is that acceptable?

Best,
Aniket

Alexandros Stamatakis

unread,
Oct 7, 2014, 1:30:23 AM10/7/14
to ra...@googlegroups.com
> About my first question -- so, I am doing it right, correct? Those are the
> bootstrap values.

yes, that's correct, the BS values are shown as branch labels, since BS
values actually refer to branches (bipartitions) of the tree and not nodes,

alexis

> About the alignment: it is a plant nuclear protein coding sequence (DNA).
> I understand that the alignment might be generating conflicting signals but
> 1/100 support is a little uneasy in any analysis. If it were at least a few
> more, it would have made sense, but here it looks like only one out of
> hundred trees support that node.

could it be that your alignment contains rogue taxa?

alexis

>
> Does that help?
>
> Best,
> Nick
>
> On 6 October 2014 11:58, William Gearty <willg...@gmail.com> wrote:
>
>> You received this message because you are subscribed to a topic in the
>> Google Groups "raxml" group.
>> To unsubscribe from this topic, visit
>> https://groups.google.com/d/topic/raxml/ir9c3vYVIWs/unsubscribe.
>> To unsubscribe from this group and all its topics, send an email to
>> 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

Alexandros Stamatakis

unread,
Oct 7, 2014, 1:32:11 AM10/7/14
to ra...@googlegroups.com


On 10/07/2014 12:44 AM, William Gearty wrote:
> I believe those are the relative node ages. I don't know too much about how
> RAxML calculates those, but I assume that those values are just summations
> of the branch lengths.

that's not correct, RAxML doesn't display this kind of information,
could you send us a screenshot of the tree?

alexis
>>>>> --
>>>>> You received this message because you are subscribed to a topic in the
>>>>> Google Groups "raxml" group.
>>>>> To unsubscribe from this topic, visit
>>>>> https://groups.google.com/d/topic/raxml/ir9c3vYVIWs/unsubscribe.
>>>>> To unsubscribe from this group and all its topics, send an email to
>>>>> raxml+un...@googlegroups.com.
>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> Aniket Sengupta,
>>>> Doctoral Aspirant,
>>>> The Hileman Lab of Plant Evolution and Development,
>>>> The Department of Ecology and Evolutionary Biology,
>>>> The University of Kansas,
>>>> Haworth Hall, Room 8022,
>>>> 1200 Sunnyside Avenue,
>>>> Lawrence, Kansas 66045.
>>>> anikets...@gmail.com
>>>> ----------------------------------------------
>>>> <^v^> *What immortal hand or eye / **Could frame thy fearful
>>>> symmetry?* -William Blake
>>>>
>>>> --
>>>> 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.
>>>> For more options, visit https://groups.google.com/d/optout.
>>>>
>>>
>>>
>>>
>>> --
>>> William Gearty
>>> people.stanford.edu/wgearty
>>>
>>> --
>>> You received this message because you are subscribed to a topic in the
>>> Google Groups "raxml" group.
>>> To unsubscribe from this topic, visit
>>> https://groups.google.com/d/topic/raxml/ir9c3vYVIWs/unsubscribe.
>>> To unsubscribe from this group and all its topics, send an email to
>>> raxml+un...@googlegroups.com.
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
>>
>>
>> --
>> Aniket Sengupta,
>> Doctoral Aspirant,
>> The Hileman Lab of Plant Evolution and Development,
>> The Department of Ecology and Evolutionary Biology,
>> The University of Kansas,
>> Haworth Hall, Room 8022,
>> 1200 Sunnyside Avenue,
>> Lawrence, Kansas 66045.
>> anikets...@gmail.com
>> ----------------------------------------------
>> <^v^> *What immortal hand or eye / **Could frame thy fearful symmetry?*
>> -William Blake
>>
>> --
>> 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.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>
>

--
Message has been deleted

Alexandros Stamatakis

unread,
Oct 7, 2014, 10:42:05 AM10/7/14
to ra...@googlegroups.com
this looks weird, it might be that these are actually normal branch
lengths, but I don't know why they are displayed like this,

alexis

On 10/07/2014 04:01 PM, Aniket Sengupta wrote:
> Hi,
>
> It is possible that there might be rogue taxa. I looking into small part of
> the CDS of a single gene across flowering plants.
>
> I am attaching screenshots of the bipartition file. In one I displayed
> "node ages" under node labels in figtree, the other has "labels" displayed.
>
> Thanks,
> Nick
>
> On 7 October 2014 00:32, Alexandros Stamatakis <

Ingo Michalak

unread,
Oct 7, 2014, 11:00:58 AM10/7/14
to ra...@googlegroups.com
Hello together,
these values only make sense for ultrametric trees (figtree was designed for BEAST, in the first place). I guess (!) in case of phylograms figtree takes the mean lengths of sister-branches and sums these going back towards the root... 

Cheers!
Message has been deleted
Message has been deleted

YL L

unread,
Oct 12, 2014, 12:32:28 AM10/12/14
to ra...@googlegroups.com
Hi , Nick, 
you need to click the "Branch Lable" and displays the "lable" for the bootstrap support values.
Cheers,
Lee



在 2014年10月7日星期二UTC+8下午10时01分34秒,Nick写道:
Hi,

It is possible that there might be rogue taxa. I looking into small part of the CDS of a single gene across flowering plants.

I am attaching screenshots of the bipartition file. In one I displayed  "node ages" under node labels in figtree, the other has "labels" displayed.

Thanks,
Nick
On 7 October 2014 00:32, Alexandros Stamatakis <alexandros...@gmail.com> wrote:



--
anikets...@gmail.com
----------------------------------------------
Reply all
Reply to author
Forward
0 new messages