[Bioperl-l] graphing trees

[Sean Davis]

On Nov 5, 2004, at 8:50 AM, Guillaume Rousse wrote:

> Sean Davis wrote:
>> On Nov 5, 2004, at 7:53 AM, Guillaume Rousse wrote:
>>> Jason Stajich wrote:
>>>
>>>> On Nov 3, 2004, at 8:39 AM, Guillaume Rousse wrote:
>>>>
>>>>> Hello.
>>>>>
>>>>> I have large trees to graph, the same way as phylogenetic trees 
>>>>> are usually drawn, with edges length expressing distances between 
>>>>> nodes.
>>>>>
>>>> Bio::TreeIO::svggraph generates SVG.
>>>
>>> Seems the most interesting right now.
>>>
>>> Actually, I currently build my tree by first computing a distance 
>>> matrix from my raw data, passing it to the Algorithm-Cluster module, 
>>> and building nodes from its ouput. So the most logical way would be 
>>> to implement a new Bio::TreeIO::cluster module, constructing a 
>>> Bio::Tree object from this output, and using Bio::TreeIO::svggraph 
>>> to output my graph. Would such piece of code eventually interest 
>>> bioperl maintainers ?
>>>
>>> Alternatively, if there was a way to construct the nodes list from 
>>> the matrix distance directly in BioPerl, so as to bypass 
>>> Algorithm-Cluster, I could also get interested. I am just effraid 
>>> that with a 3000 x 3000 matrix, a pure-perl implementation of the 
>>> clustering could have perfs issues.
>>>
>> Have you thought using R for the calculations/display?  If you have 
>> that much data, it might be worthwhile working with it in a 
>> computation environment.
> Is R usable from a perl program, and not in interactive way only ? And 
> this would only be a replacement for Algorithm-Cluster, which works 
> quite well (it is a wrapper around a C library), that wouldn't get me 
> rid of the task of building the Bio::Tree Object manually. I was just 
> asking if a clustering algorithm was available from bioperl itself.
>

As for what R can do, you can look at and do a search on the page for 
"phylo" or something like that....

http://cran.us.r-project.org/src/contrib/PACKAGES.html

As for combining with perl, there is RSperl that is fairly complex (I 
haven't gotten it to build).  There is also statistics-R 
(http://search.cpan.org/~gmpassos/Statistics-R-0.02/).  Folks have used 
R with 2-way pipes.  Finally, you can just use system calls to run R in 
batch mode (from perl, create a script file, run R->save results, read 
results).

Let me say that I have not used R for phylogenetics of any kind, but I 
do use it for microarray data, etc. and find it a very nice addition to 
my own toolbox, even for perl-like data manipulation.

Sean