Missing UniProtKB prefixes on some manually edited extensions

[jseager7]

(References https://github.com/pombase/canto/issues/2255)

One of the syntax errors in our GAF export was due to the fact that UniProtKB accession numbers were not prefixed with 'UniProtKB:' when used in annotation extensions. This has since been fixed in https://github.com/pombase/canto/issues/2363, but there are a few remaining annotations to which this change has not applied, presumably since they were edited as text.

Both annotations are in PMID:30459352. The UniProtKB identifier F9X6V3_ZYMTI is invalid, and should instead be UniProtKB:F9X6V3.

UniProtKB	Gene name	Term ID	Annotation Extension
F9X6V3	MYCGRDRAFT_91615	GO:0008897	has_direct_input(F9X6V3_ZYMTI)
F9XD62	MYCGRDRAFT_93828	GO:0140110	has_input(F9X6V3_ZYMTI),part_of(GO:0048023)

@CuzickA If the has_input and has_direct_input extensions are not available on these GO terms, you can probably fix this by editing as text and prefixing each accession number with 'UniProtKB:'. So the extensions would change as follows:

invalid	valid
has_direct_input(F9X6V3_ZYMTI)	has_direct_input(UniProtKB:F9X6V3)
has_input(F9X6V3_ZYMTI),part_of(GO:0048023)	has_input(UniProtKB:F9X6V3),part_of(GO:0048023)

If you'd prefer to have the extensions enabled, let me know and I can look at changing the relevant extension configuration.

[kimrutherford]

The code currently will only add the database prefix to identifiers of proteins that have been added to the session and it only checks the accession number. I suspect the accession number is required in extensions for GAF submission.

In this case:

has_direct_input(F9X6V3_ZYMTI) I think the ID should be "F9X6V3"

[ValWood]

That'r right. F9X6V3_ZYMTI is an identifier not an accession.

[jseager7]

That's right. F9X6V3_ZYMTI is an identifier not an accession.

Thanks for catching that. I'll update my previous comment.

[CuzickA]

I've made the change to