[Pipet Devel] Re: phylogenetic gui using pygnome

[J.W. Bizzaro]

Hi Rick!

Rick Ree wrote:
> 
> Hi, I noticed your openlab entry on freshmeat the other day, and also that
> the pygtk tools website is hosted by you.  I've been working on some
> graphical ui tools for viewing/manipulating phylogenetic trees using
> python and pygtk/pygnome, and have an almost-beta prototype application
> that allows click-and-drag branch swapping, branch rotation, etc. a la
> MacClade.  It's more proof-of-concept than anything else.  Are any of your
> group interested in phylogenies?

Certainly.  We are actually trying to implement phylogenetics into Loci.  We'll
be defining an XML/DTD for it too.

The two guys on the Loci team who have the most interest in phylogenetics are
Thomas Sicheritz (Thomas.Sicheritz at molbio.uu.se) and Tim Triche
(jabbo at mindless.com).  At the very bottom of this message is a reply to Tim from
Thomas.  This may give you some idea about Thomas's interest.

> I've just briefly perused the TULIP/Loci
> list archives, and it seems like a lot of hardcore molecular/structural
> biochem stuff, but nothing regarding phylogenies.

I know someone who would say we talk too much about sequence and not enough
about structure :-)  We're trying to cover all the bases.  And I mean all. 
Although the "core" of Loci will be very basic, the design allows for
practically anything to be added.  (Phylogeny support will be in the core.)

> I believe in the open source model, especially when it comes to scientific
> software toolkits, and started on this project after being fed up with
> closed-source phylogenetic applications that only ran on Macs and/or
> Windows.

I know what you mean.  Try finding anything like DNASTAR or MacVector for UNIX.

> Unfortunately, being a grad student in organismal biology, I
> don't have huge amounts of time to spend writing code.

Just about everyone who is now helping with Loci has told me the same thing ;-)

> But, if you have
> any interest in adding my (alpha-level) python package to your projects
> list, let me know!

Absolutely!

You're welcome to have your own account and Web site for your program.

If you want your code to be used in Loci, we should still put it in a separate
place while both are being developed (do you want a cvs module?).  Since Loci
won't include any stand-alone bioinfo GUI apps (it's hard to explain right now),
we will have to look into making a Loci version of your program later.

BTW, where can we go to check it out?

If you would like to be involved in Loci, please join the mailing list
(pipet-devel):

    http://bioinformatics.org/mailman/listinfo/pipet-devel/

And here is the list for The Open Lab:

    http://bioinformatics.org/mailman/listinfo/theopenlab/

I'll post this message to both lists.

(message from Thomas below!)

Cheers,
Jeff
-- 
J.W. Bizzaro                  mailto:bizzaro at bc.edu
Boston College Chemistry      http://www.uml.edu/Dept/Chem/Bizzaro/
--


Tim writes:
 > 
http://www.inxight.com/Inxight_Corporate_Web_Site/Edu_Org_Program/Intro_to_Program.html
 > 
 > Take a look at this tool...  looks like it could be useful for browsing
 > phylogenetic trees.

Hmm ... :-) ... it looks quite fun ... (look at the spider phylogeny)
We could adapt the rotating/zooming idea to the python phylogentic tool
(pyphy or phypy .. or physpampy ?) 
Actually I really like the idea ... phylogenetic reconstructions tend to get 
large amounts of taxa - which is not easy to see in one window.
I have to write a treeviewer module for my (hopefully) last bigger project
(phylogenomics) in my thesis. I thought I would hack it in Tcl/Tk but with
some help from other loci'ers I could try it in python.

There is no good treeviewing program for all platforms (read: nothing for
Linux and Solaris which doesn't need 8bpp color mode)
I always had some problems to code treeparsing scripts and beeing able to
represent them in a "good" way on the screen (trifurcation, distances etc.)
I could need some help here ... 

I started on some smaller versions where the branches or taxa labels (in my
case SWISSPROT ID's) are linked to yank (sequence retrieval), SWISSPROT
database, blast and clustalw - which should be connected/linked from the
whole genome map/sequence ... that seems to fit perfectly into the LOCI
way of thinking.