A few minor corrections #9
Conversation
This example was meant to show how ID tags are not needed for features without children. The example removed the ID entries for the exons, but not for the CDSs.
| ctg123 . CDS 1201 1500 . + 0 ID=cds00002;Parent=mRNA00002 | ||
| ctg123 . CDS 5000 5500 . + 0 ID=cds00002;Parent=mRNA00002 | ||
| ctg123 . CDS 7000 7600 . + 0 ID=cds00002;Parent=mRNA00002 | ||
| ctg123 . CDS 3301 3902 . + 0 ID=cds00003;Parent=mRNA00003 |
There was a problem hiding this comment.
Without the ID's here we won't be able to keep these two alternative CDS's separate, since they are both in the same mRNA.
There was a problem hiding this comment.
I don't quite understand. The parent mRNAs are different (mRNA00001, mRNA00002, and mRNA00003).
There was a problem hiding this comment.
Yeah but cds00003 and cds00004 both happen to be on the same mRNA in this example (mRNA00003).
There was a problem hiding this comment.
Ah, I see. With cds00003 and cds00004 referring to different proteins translated from the same mRNA (different start codons). OK, you're right.
The CDS ID should refer to the common ID shared by all genomic intervals in the CDS. This makes sense. So the CDS is considered a discontinuous feature? If this is the right way to do it then pretty much everyone is doing it wrong, giving every CDS row a unique ID.
There was a problem hiding this comment.
Actually in my limited experience (NCBI and Ensembl downloads for human and mouse), they seem to be doing it "right", giving all segments of the discontinuous CDS feature the same ID.
There was a problem hiding this comment.
That's good. Phytozome10, where I get most of my plant genomes, forces unique IDs. Incidentally, I think a group at my university is planning some giant analysis of GFF files, as part of a test of a new domain specific language they are designing. Maybe I can get them look into conventions in the "wild".
But anyway, I agree about the CDS IDs here. I'll push a commit in a bit that reverts the change.
There was a problem hiding this comment.
Since a CDS is discontinuous, are their IDs strictly required, even in mRNAs that contain only a single CDS?
There was a problem hiding this comment.
I'm not sure. Certainly not all CDSs are discontinuous, so I'd guess those would have even less use for an ID. For the discontinuous case I would probably personally still want to see all the chunks having the same ID just for explicitness, but I'm not sure the spec should mandate that.
There was a problem hiding this comment.
So perhaps CDS IDs should be strictly required only when multiple CDS share one mRNA.
When CDS IDs are used, should it be strictly required that each interval in the CDS share the same ID? For example, is this incorrect:
Chr1 Araport11 gene 3631 5899 . + . ID=AT1G01010;Name=AT1G01010;Note=NAC domain containing protein 1
Chr1 Araport11 mRNA 3631 5899 . + . ID=AT1G01010.1;Parent=AT1G01010;Note=NAC domain containing protein 1
Chr1 Araport11 five_prime_UTR 3631 3759 . + . ID=AT1G01010:five_prime_UTR:1;Parent=AT1G01010.1
Chr1 Araport11 exon 3631 3913 . + . ID=AT1G01010:exon:1;Parent=AT1G01010.1
Chr1 Araport11 CDS 3760 3913 . + 0 ID=AT1G01010:CDS:1;Parent=AT1G01010.1
Chr1 Araport11 exon 3996 4276 . + . ID=AT1G01010:exon:2;Parent=AT1G01010.1
Chr1 Araport11 CDS 3996 4276 . + 2 ID=AT1G01010:CDS:2;Parent=AT1G01010.1
Chr1 Araport11 exon 4486 4605 . + . ID=AT1G01010:exon:3;Parent=AT1G01010.1
Chr1 Araport11 CDS 4486 4605 . + 0 ID=AT1G01010:CDS:3;Parent=AT1G01010.1
Chr1 Araport11 exon 4706 5095 . + . ID=AT1G01010:exon:4;Parent=AT1G01010.1
Chr1 Araport11 CDS 4706 5095 . + 0 ID=AT1G01010:CDS:4;Parent=AT1G01010.1
Chr1 Araport11 exon 5174 5326 . + . ID=AT1G01010:exon:5;Parent=AT1G01010.1
Chr1 Araport11 CDS 5174 5326 . + 0 ID=AT1G01010:CDS:5;Parent=AT1G01010.1
Chr1 Araport11 CDS 5439 5630 . + 0 ID=AT1G01010:CDS:6;Parent=AT1G01010.1
Chr1 Araport11 exon 5439 5899 . + . ID=AT1G01010:exon:6;Parent=AT1G01010.1
Chr1 Araport11 three_prime_UTR 5631 5899 . + . ID=AT1G01010:three_prime_UTR:1;Parent=AT1G01010.1
There was a problem hiding this comment.
Maybe I can get them look into conventions in the "wild".
My fantasy: a registry of all GFF3 files in the wild (or at least a start with a representative sample), and a Jenkins or other CI job that regularly runs a GFF3 validator and reports variations from normative standard and de-facto conventions....
@nathandunn how hard would this be?
alexhenrie
force-pushed
the
master
branch
6 times, most recently
from
6f5195d to
fe73505
Compare
Just a few minor corrections.