[Bioperl-l] Phyloxml <alignment>

Arlin Stoltzfus arlin.stoltzfus at nist.gov
Wed Jul 30 09:58:09 EDT 2008


Jason, et al--

In regard to the Bio::NEXUS objects, we (Bio::NEXUS team) have the  
aspiration to integrate with BioPerl objects via some coordinated  
CDAT (character-data-and-trees object) super-object that would allow  
coordinated operations like slicing OTUs X, Y and Z out of both the  
tree and the alignment.

But it hasn't really happened.  Aaron suggests that there might be a  
lesson here, and maybe its about the folly of designing one great  
object ("One Thing to rule them all; One Thing to find them; One  
Thing to gather all, and in the darkness, bind them").  Anyway, Aaron  
suggests we should opt for a "dumber" (his words) approach such as  
treating an alignment as an "annotation" of a tree.  He and Weigang  
have done some of that.

Yet, I think Rutger might have worked out a smarter approach in  
Bio::Phylo, which (as I understand it) *does* have bindings to  
BioPerl objects.

It might be worth having a summit meeting to work out some strategy  
to solve this problem.  We might be able to get support for a meeting  
or hackathon via NESCent if we write a White Paper that lays out a  
plan aimed at 1) object bindings between Bio::Phylo, BioPerl and  
Bio::NEXUS and 2) some coordinated operations, with or without a  
super-object.  The purpose would be to expand the power of these Perl  
libraries to facilitate evolutionary analyses.

Any takers?

Arlin

On Jul 29, 2008, at 12:45 PM, Jason Stajich wrote:

> Mira -
>
> I've not thought too clearly about how this should go, certainly  
> Bio::AlignIO should be reused for the parsing aspect, although I'm  
> not clear what format the alignment is in?
>
> At some point this is might need to be unified with the NEXUS-type  
> objects that the Bio::NEXUS group has worked on.  We may have to  
> work with them to define the proper object here that holds both  
> tree and alignment data and allows slicing out either from the object.
>
> So a super-object is probably going to have to be created to  
> marshall out the trees and alignment data.  I didn't realize the  
> spec was going as far as to encapsulate all of the alignment data  
> as well.  We may need to think about whether we can support all of  
> the aspects of the spec at this time as well.
>
> I hope other people have input here as well.
>
> -jason
>
> On Jul 29, 2008, at 9:20 AM, Han, Mira wrote:
>
>>
>> Hi,
>> Sorry for bombarding the list
>> Thought this would be better on a separate thread.
>> Phyloxml documents have <phylogeny></phylogeny>s.
>> But also <align:alignment></align:alignment>s.
>> What would be the standard approach for parsing,
>> I guess TreeIO->next_tree
>> shouldn't take care of the alignments.
>> Would I need to make a AlignIO->next_aln for the phyloxml document  
>> as well?
>>
>> Mira
>>
>>
>> _______________________________________________
>> Bioperl-l mailing list
>> Bioperl-l at lists.open-bio.org
>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
>
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/bioperl-l

------------
Arlin Stoltzfus (arlin.stoltzfus at nist.gov), Research Biologist, NIST
Fellow, CARB, 9600 Gudelsky Drive, Rockville, Maryland  20850
tel 240 314 6208, fax 240 314 6255, www.molevol.org/camel




More information about the Bioperl-l mailing list