[phylocom-user] phylocom-user Digest, Vol 2, Issue 3

Steve Kembel skembel at berkeley.edu
Tue Aug 19 12:10:52 PDT 2008


Hi Juan, Jenny,

OK, so if I understand you correctly, Juan, your problem is not  
actually that the sampleprune function itself is giving a NA error,  
it's that the trees you generate with sampleprune give you NA values  
when you run aot?

I suspect the problem you're both having is due to zero-length  
branches in your tree. If there are branches of length zero, when you  
run aot the standardized contrast at each node is calculated by  
dividing the raw contrast by the branch length, which is a division by  
zero error. And since the measure of signal is based on the  
standardized contrasts, you're also getting NA's in the aotf signal  
output.

There are two ways zero-length branches can come about. One is easy to  
fix - if there are any branches of length 0.00 in the tree file,  
change that to some small non-zero value like 0.001. Consider  
collapsing zero length or very short branches to polytomies using the  
ape package in R or similar.

The second is a bit trickier and is caused by missing branch lengths  
in the tree. If there are any nodes with no branch length listed in  
the phylo file, phylocom assumes the branch length on that node is  
zero. The most common way for this problem to appear is when your tree  
has a 'tail' - an outer set of parentheses with no branch length  
listed. The way to fix this is either to remove the tail, which in a  
Newick tree string is the outermost set of parentheses, or to add some  
small non-zero branch length to the tail. Look at your phylo file. If  
it looks like something like this:
(((A:3,B:3):7,C:10):5);
...the outermost set of parentheses with no ":" after it is the tail.  
Delete those parentheses and your aot analyses should work with no NAs  
in the output. So after deleting the tail your tree would now look  
like this:
((A:3,B:3):7,C:10):5;

Cheers,
Steve

On Aug 19, 2008, at 11:03 AM, Juan Pablo Gomez Echeverri wrote:

> Hi
>
> I am having the same problem as you are.  I have tried with several  
> phylogeny and traits files but the only way it worked for me was to  
> use the exact output of the phylogeny form RAxML without prunning  
> the phylogeny to the community list.  I am trying to solve this  
> problem too.  Please let me know if you figure something out. I don 
> ´t know if it is a problem with branch length format or something  
> like that.
>>
>> From: Jenny Read <jenny.read at sci.monash.edu.au>
>> Subject: [phylocom-user] aot problem
>> To: phylocom-user at lists.phylodiversity.net
>> Message-ID: <48A9F76C.8080700 at sci.monash.edu.au>
>> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>>
>> Hi,
>>
>> We are having a problem getting AOT to work (using Windows) in  
>> Phylocom
>> v. 4.  The output gives the numbers of traits, taxa, contrasts, but  
>> does
>> not give the VarCn or PicR ('NaN' is given).  I have no trouble  
>> getting
>> the example files to work, so I assume the problem is something to do
>> with my file format (saved as text, tab-delimited, then renamed  
>> without
>> the extension).  We have tried quite a few variations on file
>> formatting, but with no luck.  Can anyone suggest what we are doing
>> wrong please ?
>>
>> Many thanks,
>> Jenny.
>>
>> ***
>
> Juan Pablo Gomez



More information about the phylocom-user mailing list