gff3 specification

[nhartwic]

The gff-version listed by helixer is "3.2.1" I haven't been able to find a specification for that version. The only particularly well detailed specification of gff3 that I've ever found is from Sequence Ontology for "3.1.26"...

https://github.com/The-Sequence-Ontology/Specifications/blob/master/gff3.md

...According to this standard, CDS lines that are meant to be part of the same sequence should have the same ID. This marks those lines as a multi-line feature. Helixer doesn't do this. Helixer marks each CDS segment with its own ID.

Is there an actual specification for 3.2.1 somewhere? If so, does it disagree with 3.1.26 in terms of how CDS are supposed to be represented?

EDIT: For reference, current GFF from helixer looks like...

1       Helixer gene    241717  242876  .       +       .       ID=Zmays.B73.HPIv02_1_000003
1       Helixer mRNA    241717  242876  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1;Parent=Zmays.B73.HPIv02_1_000003
1       Helixer exon    241717  241720  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1.exon.1;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer five_prime_UTR  241717  241717  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1.five_prime_UTR.1;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer CDS     241718  241720  .       +       0       ID=Zmays.B73.HPIv02_1_000003.1.CDS.1;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer exon    241835  242876  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1.exon.2;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer CDS     241835  242875  .       +       0       ID=Zmays.B73.HPIv02_1_000003.1.CDS.2;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer three_prime_UTR 242876  242876  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1.three_prime_UTR.1;Parent=Zmays.B73.HPIv02_1_000003.1

...According to the sequence ontology standard, it should be...

1       Helixer gene    241717  242876  .       +       .       ID=Zmays.B73.HPIv02_1_000003
1       Helixer mRNA    241717  242876  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1;Parent=Zmays.B73.HPIv02_1_000003
1       Helixer exon    241717  241720  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1.exon.1;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer five_prime_UTR  241717  241717  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1.five_prime_UTR.1;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer CDS     241718  241720  .       +       0       ID=Zmays.B73.HPIv02_1_000003.1.CDS;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer exon    241835  242876  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1.exon.2;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer CDS     241835  242875  .       +       0       ID=Zmays.B73.HPIv02_1_000003.1.CDS;Parent=Zmays.B73.HPIv02_1_000003.1
1       Helixer three_prime_UTR 242876  242876  .       +       .       ID=Zmays.B73.HPIv02_1_000003.1.three_prime_UTR.1;Parent=Zmays.B73.HPIv02_1_000003.1

I know the difference seems small, but its the difference between a "correct" parsing of the file producing two distinct proteins and producing a single (correct) protein.

[alisandra]

Ah, thanks for raising this. That will help us get it patched up the next round of changes / next version.

As a practical option until then, many parsers use the Parent feature to determine what belongs together and will parse this correctly despite the IDs. I think gffread should do it.