Re: [SO-devel] GFF3 Specification

[Barry M. <bar...@ge...>]

Thanks Peter for the useful feedback.

The reference to the Order tag is a mistake - it's text that snuck in somehow from another version of the document where we were exploring the possible solutions.  I'll take that out now.

In the case of the multiple ##sequence-region directives for the same seqid.  That was a clarification that I added thinking that it added clarity to the text without breaking backwards compatibility, but perhaps I was wrong.  In the example you describe why wouldn't you just pull out the the features that you want, but still allow the ##sequence-region to specify the full range?

This:

##sequence-region chr1 10000 52000

Instead of this:

> ##sequence-region chr1 10000 12000
> ##sequence-region chr1 50000 52000


The second form seems to add significant complexity to the parser without obvious benefit to me.  You now have to range overlap all features against all ##sequence-region directives for a given seqid to validate.  However, if you broke the chromosome into parts (in the fasta file or ##FASTA section), you would have to use different seqids for the separate sequences and thus you'd have to adapt the sequence-region directives accordingly.

For example, if you don't split the actual sequences then having the sequence-region span the full range of the sequence (or at least the full range of the annotated features), seems simpler and sufficient.

##sequence-region chr1 1 .. 52000
…
##FASTA
>chr1
ACGT…TGCA

----------------------

If you are splitting out the sequences themselves then you have to re-name and the coordinate ranges change.

##sequence-region chr1_partA 1 .. 12000
# The next one is really 50000-52000
##sequence-region chr1_partB 1 .. 2000
…
##FASTA
>chr1_partA
ACGT…TGCA
>chr1_partB
ACGT…TGCA

On Feb 26, 2013, at 11:32 AM, Peter Cock wrote:

> On Tue, Feb 26, 2013 at 5:02 PM, Barry Moore <bar...@gm...> wrote:
>> Hi All,
>> 
>> The GFF3 specification has been updated to version 1.21.  Changes are minor
>> - mostly involve clarification to the wording.  Changes include the
>> following:
>> 
>> • Clarification of escaping conventions.
>> • Explicit requirement that the value of start and end be one-based positive
>> integers.
>> • Clarification to the use of quotes in attribute values.
>> • Clarification of lines begining with # and exclusion of inline comments.
>> • Clarification that the ##gff-version pragma only appears once in a file.
>> • Clarification to the ##sequence-region pragma.
>> 
>> The current specification can be found at:
>> http://www.sequenceontology.org/resources/gff3.html
>> 
>> If you would like to review the changes in version 1.21 they are highlighted
>> in the document:
>> http://www.sequenceontology.org/resources/gff3_1.21.html
>> 
>> The first meeting of the GFF3 working group was held on February 7th and
>> discussed how to annotate partial features and circular genomes.  This work
>> is ongoing and will ultimately result in proposed updates to the
>> specification and/or best practices guidelines.
> 
> I noticed to things I wanted to bring up.
> 
> There appears to be a problem in the current text on features spanning
> the origin of a circular genome referencing the new proposed "Order" tag,
> yet that tag is not listed (yet). As an aside, I personally prefer the name
> "Part" as in Part=1/3, Part=2/3 and Part=3/3, but the term isn't critical.
> 
> Separately, the new text forbids multiple ##sequence-region lines
> for the same seqid. Was the use-case of describing two (or more)
> sub-regions of a large reference rejected? For example, extracting
> a reduced GFF3 file to include just the locus of two genes of interest,
> say:
> 
> ##sequence-region chr1 10000 12000
> ##sequence-region chr1 50000 52000
> 
> Regards,
> 
> Peter
> 
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_feb
> _______________________________________________
> SOng-devel mailing list
> SOn...@li...
> https://lists.sourceforge.net/lists/listinfo/song-devel

Barry Moore
Research Scientist
Dept. of Human Genetics
University of Utah
Salt Lake City, UT 84112
--------------------------------------------
(801) 585-3543