[Bioperl-l] Re: [SO-devel] GFF3 - Bioperl - SO

[Steffen Grossmann]

Allen,

it might be understandable that you don't want to use the tag/value
annotation system, but it is then problematic to let
Bio::SeqFeature::Annotated inherit from Bio::SeqFeatureI which requires
to implement the tag/value scheme. This is some kind of indirect
deprecation of functionality, which I don't think helps in making things
clearer.
Moreover, Bio::FeatureIO::gff is only able to write out features of the
type Bio::SeqFeature::Annotated, so I think that the discussion about
deprecating Bio::Tools::GFF and replacing it with Bio::FeatureIO::gff is
dangerous, since Bio::SeqFeature::Generic is of widespread use and GFF
output of such features should be maintained.

These are just some thoughts about the structure, I'd be curious to
learn about your and other people's opinions about it.

By the way, also Bio::SeqFeature::AnnotationAdaptor does not work
together with Bio::SeqFeature::Annotated, so the workaround proposed
below does not work.

Steffen

Allen Day wrote:

>On Sun, 17 Oct 2004, Steffen Grossmann wrote:
>
>  
>
>>Ok, ok, if you really want to convince me that Bio::FeatureIO::gff is 
>>the technology of the future... (although I would also like to hear 
>>Chris Mungall's opinion on this point...)
>>    
>>
>
>I'd like to hear Chris weigh in as well.
>
>  
>
>>So let's get back to the technicalities.
>>
>>There is something which prevents the use of 
>>Bio::SeqFeature::Tools::IDHandler with Bio::FeatureIO::gff, namely the 
>>extensive use of the tag/value annotation system in the former. On one 
>>hand, this comes from repeated calls of Bio::SeqFeatureI's 'primary_id' 
>>method, on the other hand there are calls of 'get_tagset_values'.
>>    
>>
>
>We can map Bio::SeqFeature::Annotated::primary_id to something reasonable
>in the AnnotationCollection, but my guess is that avoiding the tagset will
>be more difficult.
>
>I'd prefer not to add SeqFeatureI compliance as far as tagset methods go
>-- one of the main reasons I added SeqFeature::Annotated was to get rid of
>this "feature".  I haven't looked at the AnnotationAdaptor you mention 
>below, but it sounds like a reasonable approach.
>
>Regarding your earlier mail about returning unflattened features, we can 
>do this pretty easily by implementing GFF3 '###' tag handling.  I don't 
>remember if I mentioned this before.
>
>-Allen
>
>  
>
>>I am not completely sure, but I think that compliance of 
>>Bio::SeqFeature::Tools::IDHandler with both concepts can be achieved by 
>>two changes:
>>
>>1) Bio::SeqFeatureI::primary_id and Bio::SeqFeature::Annotated have to 
>>be made compatible. Just create a Bio::SeqFeature::Annotated object and 
>>try to call 'primary_id' on it to see that it isn't at the moment. In 
>>principle this can be done by storing the ID in the 
>>AnnotationCollectionI object when the tag/value system is not present. 
>>The problem is that the tag/value system is present in 
>>Bio::SeqFeature::Annotated (it has to be, because it implements 
>>Bio::SeqFeatureI), but it is not functional. Any suggestions?
>>
>>2) Bio::SeqFeature::Tools::IDHandler can be made compatible with both 
>>annotation systems by using the Bio::SeqFeature::AnnotationAdaptor 
>>module. This should  not cause any problems...
>>
>>Steffen
>>
>> 
>>
>>Allen Day wrote:
>>
>>    
>>
>>>On Thu, 14 Oct 2004, Steffen Grossmann wrote:
>>>
>>> 
>>>
>>>      
>>>
>>>>Dear Allen,
>>>>
>>>>I meanwhile understood that Bio::Tools::GFF in connection with 
>>>>Bio::SeqFeature::Tools::IDHandler is doing a lot of the stuff I'd like 
>>>>to have. Somehow, Bio::FeatureIO::gff seems to be a parallel development 
>>>>to the first alternative. I don't know in how far there are plans to 
>>>>   
>>>>
>>>>        
>>>>
>>>I don't see from a quick look at the source how Bio::Tools::GFF is related
>>>to Bio::SeqFeature::Tools::IDHandler.  There's nothing preventing use of
>>>the IDHandler in Bio::FeatureIO::gff, in fact it sounds like what you've
>>>proposed to add so part of your work is already done.
>>>
>>> 
>>>
>>>      
>>>
>>>>bring the two approaches together, but at the moment that seems to be 
>>>>more complicated than just bringing one approach to an acceptable state. 
>>>>Since the Bio::Tools::GFF approach seems to be ahead, I will focus on it 
>>>>at the moment.
>>>>   
>>>>
>>>>        
>>>>
>>>I would advise against adding more features into Bio::Tools::GFF.  I can't
>>>speak for all others, but my future development will not use it, and I'm
>>>in the process of converting code which does use it to depend on
>>>Bio::FeatureIO::gff.
>>>
>>>-Allen
>>>
>>> 
>>>
>>>      
>>>
>>
>>    
>>
>_______________________________________________
>Bioperl-l mailing list
>Bioperl-l at portal.open-bio.org
>http://portal.open-bio.org/mailman/listinfo/bioperl-l
>
>  
>


-- 
%---------------------------------------------%
%            Steffen Grossmann                %
%                                             %
% Max Planck Institute for Molecular Genetics %
%      Computational Molecular Biology        %
%---------------------------------------------%
%              Ihnestrasse 73                 %
%               14195 Berlin                  %
%                 Germany                     %
%---------------------------------------------%
%         Tel: (++49 +30) 8413-1167           %
%         Fax: (++49 +30) 8413-1152           %
%---------------------------------------------%