[BioPerl] Re: [Bioperl-l] gff_string on an HSPI object is not Bio::DB::GFF friendly

[Jason Stajich]

Mark -

Bio::Tools::GFF is really dumb it assumes what you give it is what you
want to print.

It just takes whatever seqfeature data you pass it and tries to make it
GFF, it is not SO aware, attempt to validate, or anything else.  You have
to clear out the extra tag/values you don't want before sending it to this
writer - this is what I do in search2gff.  I don't expect a Search::HSP
object to be able to just be chunked through Tools::GFF automatically at
this point - the Search stuff was developed before SO/GFF3 was described.
You really need to custom build up a new object based on the data in the
HSP before sending it to the GFF writer.

Perhaps you want an HSP-2-SO type factory or something (which is basically
what is in the script...).

As for the .. versus + -

What does:
 my @values = $feature->get_tag_values('Target');
 print join(",", at values), "\n";
give you?

If it doesn't return 3 values of gi|2828774,54232,54206 then the data is
stored improperly in the seqfeature.

-jason

On Fri, 9 Jan 2004, Mark Wilkinson wrote:

> On Fri, 2004-01-09 at 11:22, Scott Cain wrote:
>
> >   - be sure to use a SO term for the type (ie, match or one of its
> > children)
>
> So... actually the existing implementation of GFF3 in bioperl
> from Bio::Tools::GFF->new(-gff_version => 3)
> does not generate correctly formatted GFF3 for alignment features, yeah?
>
> e.g. for column 9 of an alignment feature I get:
>
> 	Target=gi|2828774:54232..54206
>
> whereas I think I should be getting
>
> 	Target=gi|2828774+54232+54206
>
> In addition, it passes through all sorts of other tags that begin with
> capital letters:
>
> 	Bits=46.1;FracId=0.962962962962963
>
> these should be
>
> 	bits=46.1;fracId=0.962962962962963
>
> if I am reading the spec correctly.
>
> Finally, the column-3 term that comes out is "similarity", but it should be
> one of the *match terms.  Is that also correct?
>
> Please confirm that I am interpreting the GFF3 spec correctly for these
> Alignment features and I would be happy to go in and fix things (a.k.a. break
> everyone else's tools ;-) )
>
> Cheerio!
>
> Mark
>
>

--
Jason Stajich
Duke University
jason at cgt.mc.duke.edu