problem with constraining trees

[elasek-ne...@mbl.edu]

I ran several RAxML analyses specifying different constraints for each
analysis using the command:

raxmlHPC -f a -s input_filename -x 123456 -N 100 -n output_filename -g
constraint_file(in newick format) -o outgroup_names -m PROTOGAMMAMTREV

In the information file pertaining to each analysis, I received the
following message:

"The topologies of all Bootstrap and ML trees will adhere to the
constraint tree specified in constraint_filename,"

suggesting that RAxML could read my constraint file.

However, I received the same results for each analysis. In other
words, none of my trees were constrained.

I'm just wondering why this might have happened; is there anything
wrong with the way I set up my analysis?

Thanks,
Erica

[stamatakis]

Hi Erica,

There have been a couple of problems with constraints that should all
be fixed in the current version RAxML 728.

Would you check with 728 please or else send us the data?

Alexis


On 12 Jul., 21:28, "elasek-nesselqu...@mbl.edu" <elasek-

[elasek-ne...@mbl.edu]

Thanks, Alexis. I am running RAxML 7.2.8 now and will tell you if
that fixes the problem.

On Jul 12, 5:03 pm, stamatakis <alexandros.stamata...@gmail.com>
wrote:

[stamatakis]

Just to clarify something:

A frequent problem with understanding how multi-furcating constraint
trees in RAxML work is the following:

You may have 10 taxa, but only define an incomplete constraint not
including all taxa like this:

((T1, T2, T3, T4), T5, T6);

The way this is defined in RAxML is that the remaining taxa T7, T8,
T9, T10 can be inserted
anywhere in the tree and can thus potentially violate the monophyly of
(T1, T2, T3, T4).

If you want to make sure that T1, T2, T3, T4 are monphyletic without
anything else being inserted, you'd specify
a multi-furcating constraint including all taxa like this:

((T1, T2, T3, T4), T5, T6, T7, T8, T9, T10);

Also see the explanation of this at the bottom of the page:
http://wwwkramer.in.tum.de/exelixis/hands-On.html





On 14 Jul., 17:18, "elasek-nesselqu...@mbl.edu" <elasek-

[Alexis]

I have added some additional info explaining RAxML constraints to this
page:

http://wwwkramer.in.tum.de/exelixis/hands-On.html

Alexis

[Simey]

Hi,

I am playing with backbone constraints in RAxML. I have a concatenated 4 gene alignment and want to add a single OTU to the alignment; it is an alternate allele of one of the genes for one of the taxa (I cloned all heterozygous genes). I added only the alternate allele to the alignment (the rest of the genes in the alignment are represented by question marks (?)). I ran the following command:

raxmlHPC-PTHREADS -r 4geneRAXtree.txt -m GTRGAMMA -s Tr_4geneALIGN.phy -q Tr_parts.txt -n JFP624.txt -o tp22_OG T 8 

"4geneRAXtree.txt" is a RAxML tree with the original 45 taxa in the alignment.

I get the topology that I expect, but I am getting a warning from RAxML that "You provided an incomplete starting tree 45 alignment has 46 taxa" 

Is this something to be concerned about? 

[Alexis]

no nothing to worry about, this text is just printed to point out to
users that they have provided an incomplete starting tree,
in case that this is not what they want to do,

Alexis

[Alex SL]

Hi,

I am using RAxML 7.2.8 Blackbox on CIPRES and have the same issue - the output reads "The topologies of all Bootstrap and ML trees will adhere to the constraint tree specified", but then RAxML completely ignores my constraint, and I receive the same resulting tree as from an unconstrained analysis. I have tried various formats:

((1,2,3),4,5,6); or ((1, 2 ,3 ), 4, 5, 6); or ((1,2,3),(4,5,6);

but nothing seems to make any difference. The tree contains all taxa in the dataset, and all I want is for a defined part of them to be forced into monophyly to compare the best tree from the constrained search against the best one from an unconstrained analysis with a KH test. Is this a known issue? Or am I just making some silly mistake with the tree format? Any help would be appreciated.

Thanks in advance,
Alexander SL

[Alexis]

There are two possible explanations:

1. The RAxML version running at CIPRES has not been updated to the
latest GIT release of standard RAxML (please contact SDSC directly
about this)
2. There may be a misconception of how RAxML handles and interprets
constraint trees, please see previous discussion about this in here.
It seems to be a common problem, that users do not provide a full
constraint containing all taxa.

Alexis

[Alex SL]

Thanks for the quick reply. Unless I severely misunderstand something, it cannot be the latter, as the constraint was carefully crafted by my student to contain all the taxa in the dataset. Because time is running out, she will now have only Templeton tests in her project, but I will experiment around a bit more and if nothing helps try to contact CIPRES.

Cheers,
Alexander SL

[Alexis]

Dear Alexander,

The best way to proceed is to download and install the most recent
standard RAxML version from here:

https://github.com/stamatak/standard-RAxML

I am pretty sure that all constraint related issues are resolved in
this version.

If this still doesn't work, just send the constraint and dataset via
email directly to me and I will have a look.

Alexis

[Alex SL]

We have it on one of our own computers now, and it works as it should. Quicker than I thought we have finished all four KH tests! Thanks again,

Alexander SL

[Alexis]

:-)

Alexis