clarification on "match" terms

[murphyte]

Hi SO -- I'm looking for clarification on a couple of the "match" SO terms:

1. For "translated_nucleotide_match" (SO:0000181), is that for alignments with nucleotide in column 1, and protein as the target? That seems to match at least some common usage (e.g. WormBase), but it's not entirely clear from the SO definition.
2. The GFF3 specifications also include "nucleotide_to_protein_match", but that doesn't seem to exist in SO. Should it be added as a child of protein_match (assuming I'm correct about #1)?
3. The GFF3 specifications include "nucleotide_motif" as a type of match, but the SO term is a child under biological_region, which would seem to conflict with using that SO term to describe an alignment. Perhaps this just needs updating in the GFF3 specs?

thanks!

-Terence

[keilbeck]

Hi Terrance, We are working thru this now using the varieties of blast as examples. We will get back to you. --K

[murphyte]

Thanks Karen. One more observation I made just this morning. The GFF3 specs: https://github.com/The-Sequence-Ontology/Specifications/blob/master/gff3.md use "translated_nucleotide_match" for a TBLASTX example, implying the intent is to use it for nucleotide-nucleotide matches made via 6-way translations of both sequences. I think that's reasonable, given it's a child of nucleotide_match.

Perhaps there's a use case for nucleotide_to_protein_match and protein_to_nucleotide_match, to go with BLASTx and tBLASTn alignments (or the other way around?). Or we're thinking to just use protein_match, and leave it to documentation to say that that means the Target in column 9 is a protein. I think I'm satisfied with doing that.

That would leave all of the issues here as improvements needed for the GFF3 specs, rather than SO.

-terence