[Phylobase-devl] Phylobase GSoC idea

[Thibaut Jombart]

Hilmar Lapp wrote:
> This sounds great! -hilmar
>
> On Mar 10, 2008, at 8:38 PM, Peter Cowan wrote:
>
>   
>> Here is another idea, this one in need of a mentor.  Thibaut, Ben?
>>
>> Peter
>>
>> Rationale
>>
>> Methods for visualizing phylogenetic trees and associated data have
>> not kept pace with growth in tree or dataset size.  At a recent
>> NESCent hackathon a new R package for comparative methods, phylobase,
>> was designed.  The phylobase package uses modern R classes and methods
>> to store and manipulate phylogenetic trees.  Currently plotting of
>> phylogenetic trees in phylobase relies on the R base graphics, and
>> functions in other packages.  These implementations are not well
>> suited to displaying either large phylogenetic trees or trees with
>> large amounts of associated data.
>>
>> Approach
>>
>> The R programming has two primary plot device interfaces, the base
>> graphics interface and the newer, more extensible grid system.  The
>> grid system allows for a much more flexible system which will allow
>> for consistent scaling and resizing of trees and data.  Current plot
>> methods in phylobase are based on base graphics and suffer from
>> resizing and layout difficulties.  This project will develop new plot
>> methods based on the grid system.
>>
>> Challenges
>>
>> The primary challenge will be writing algorithms for efficiently
>> converting tree structures to the grid language.  Examples of similar
>> algorithms exist for plotting using the old base graphics interface.
>>
>> Involved toolkits or projects
>>
>> phylobase, R, C/C++, grid
>>
>> Mentors
>>
>> Thibaut?
>>
>>     
Hi all,

This project sounds nice to me, but I won't have time to act as a mentor 
(last months of PhD... arggg...).
Something I find interesting is trying to use a plot instead of a symbol 
for one species and one variable. This would be done using grid, 
definitely. The values observed for a species must be mean or median of 
a distribution most of the time, and it would be nice to display the 
whole distributions instead (e.g. by an histogram). That would also 
allow to represent qualitative variables as well (using barplots for 
instance). This could be done on reasonably small phylogenies, but it 
might still be worth it.

Cheers,

Thibaut.

-- 
######################################
Thibaut JOMBART
CNRS UMR 5558 - Laboratoire de Biométrie et Biologie Evolutive
Universite Lyon 1
43 bd du 11 novembre 1918
69622 Villeurbanne Cedex
Tél. : 04.72.43.29.35
Fax : 04.72.43.13.88
jombart at biomserv.univ-lyon1.fr
http://lbbe.univ-lyon1.fr/-Jombart-Thibaut-.html?lang=en
http://adegenet.r-forge.r-project.org/