Error running gffcompare about gffcompare HOT 3 OPEN

Really? Are you really going to argue that you just don't see a duplicate ID gene29892 in those first two lines of your ref.gff quote you're showing here ? :)

While the error message might not be very accurate (there is no actual overlap there, for those two duplicated gene IDs), the problem still stands: the ID gene29892 has obvious transcript properties (parenting CDS and exon features) while it is being declared twice in the file (1st at 74631-74744, 2nd time at 146276-147073). This is not a valid GFF3 format, as my understanding is that the feature IDs should be unique.. (though indeed my GFF parser is a bit more lenient about this, accepting non-unique IDs if the features are on separate reference sequences.. which is not the case here).

Then things went downhill ("downstream") from there.. according to the merciless GIGO principle.. I hope you understand, it's hard to somehow automagically fix invalid input reference annotation data, or guess what the authors of those data really meant there -- e.g. that it was somehow a 2-part gene there (?!), so it's OK if they broke the ID uniqueness rule on a whim..

from gffcompare.

Leaving aside the cheeky fun I had with my reply above, it turns out that in fact I was in the wrong there -- not about the duplicate IDs, but about the validity of that annotation, as this is a special case of trans-splicing where actually the current GFF3 specification does allow for the same ID in the case of discontinuous features like trans-splicing and fusions.
So please accept my belated apologies for the incorrect/incomplete answer -- and thanks for this trans-splicing example! I ran into this old closed issue while looking for trans-splicing examples so I can add trans-splicing support to my GFF parser (and thus to gffread, gffcompare etc.).

from gffcompare.

Hello Geo!

Glad to hear that. It's good that we are coming across such scenarios and I am happy that the project is on constant development. Taking it on a positive note.

Your hard work is much appreciated!

Regards
Vijay

from gffcompare.