GPAD/GPI 2.0 Specifications - Request for comments on new draft

[balhoff]

Starting on Tuesday, July 18th, we are requesting review and comments on the proposed GPAD/GPI 2.0 file format specifications. The spec has been revised since the last round of comments.

https://github.com/geneontology/go-annotation/blob/master/specs/gpad-gpi-2-0.md

Please add any comments or questions you have about the specs to this ticket by Friday, August 18th, 2023.

Thank you.

[ukemi]

I sent these to Lori and looked them over as well. You did a nice job @balhoff. But as Lori said, we will see what happens when we start doing the new loads. :)

[kltm]

Some initial comments:

• '!gpa-version: 2.0' seems a little quirky; should this not be gpad-version? I know at one point an allowed file extension was .gpa, but that is no longer allowed. Does the history justify the inconsistency?
• There is the BNF comment that complementation is written using minus symbol; do we have complements anywhere? I'm not sure I ran into that.
• Date is a little funny for us. We want YYYY-MM-DD; time may be included as YYYY-MM-DDTHH:MM:SS, so ISO-8601 parsers will work, but we are not wanting to accept all ISO-8601 (i.e. YYYYMMDD would be verboten). This is to preserve trivial parsing without worrying and the entire spec and necessitating a library.

[balhoff]

• gpad-version sounds good to me.
• I think the only place that a complement is used is in Text_No_Spaces, but really that easily could be defined by creating something like Nonspace_Text_Char
• The date definition I have requires the hyphens. I think it's better we stick to xsd:date rather than ISO-8601 (I think xsd:date is the subset of ISO we want).

[kltm]

Tagging @alexsign just to make sure this is on your radar. We're not switching to this immediately, but we are trying to firm up the spec for future conversation.

[kltm]

@balhoff

• Okay, so if there are no complaints gpa-version to gpad-version.
• Ah, I missed it there. I just wanted to make sure it was used if mentioned. Either way.
• I'm a little worried that xsd:date is a bit...marginal? Maybe that's just me. How about noting that it's a subset of ISO 8601 as defined by xsd:date (https://books.xmlschemata.org/relaxng/ch19-77041.html, not sure what the best link is for an official ref)? I suspect that most people outside reading this would be more familiar/comfortable with ISO 8601 vs xsd:date.

[sierra-moxon]

Very clear - thank you! One trivial comment - DB_Object_Synonyms uses the value in the example for DB_Object_Symbol. Should all symbols be synonyms as well?

[balhoff]

@sierra-moxon I don't know the answer to your question. Maybe @vanaukenk or @ukemi could comment? The issue is that the example values have AMOT for DB_Object_Symbol, and AMOT|KIAA1071 for DB_Object_Synonyms. I agree that does kind of lead you to a conclusion that might not be intended.

[pgaudet]

RIght, I dont think symbols should be synonyms. In some cases maybe people want to capture at least different casing? But I suggest we remove AMOT form synonyms in the examples.

Thanks,

Pascale

[balhoff]

Thanks @pgaudet, I changed the example: #4724

[kltm]

@pgaudet Not sure how/why this is in the "Misc" project and not the "GPAD/GPI 2.0" project...
Unless you can remember why, I would suggest switching it over to there and probably closing.

[pgaudet]

Right. This is anyways done.