Should we allow the use of URI syntax in SSSOM tables?

[matentzn]

To be self contained, the tsv format metadata should provide a curie map - we will make this mandatory.

[wdduncan]

I am working with the Darwin Core group on the Darwin Core to MIxS mappings. See this repo:
https://github.com/tdwg/gbwg

In their SSSOM spreadsheet, they have been using full URIs. Why do we want prohibit this?

cc @cmungall @raissameyer @pbuttigieg

[balhoff]

Should full URIs be enclosed in angle brackets, to make CURIE vs. URI obvious?

[matentzn]

Hmmmm... I dont know. Can I ask why? In the original proposal the whole point was to zone in on a terse, curie based representation for the tables that can be handled by standard python libraries.. I am for now slightly against this, but I can be convinced otherwise.

[balhoff]

Personally I find CURIEs frequently an unnecessary complication, but I haven't been much involved here so take my opinion with a grain of salt. :-)

[matentzn]

I personally agree with that as well - the amount of times I wrote curie-iri converters is uncountable :D I just feel that this is too much of burden to standardisation.. SSSOM plus some standard curie maps to refer to would allow a very easy way to compare stuff, and not bring us back into the country of UMLS cui vs OBO purl vs SCTID etc.. Lets see what @cmungall says..

[wdduncan]

We can (of course) add a section to the SSSOM document that is used to define the URI to CURIE mappings.

[matentzn]

The main idea of restricting to CURIEs is, correct me if I am wrong @cmungall, that the whole design of SSSOM revolves around the idea that the files should be easy to use for regular bioinformaticians - and URIs are considered an ugly distraction. I am super torn about the overhead for the semweb/obo community to having to provide prefixes each time they emit an sssom file..

One compromise could be that we are more permissive in sssom parse, basically allowing TSV files to be passed in that have IRIs, but spit out the proper, clean prefixed SSSOM files using the linkml curie map. What do you think @cmungall ?

[wdduncan]

FYI for the DWC-MIXS mapping we re removing quotes from the curies.

cf. DWC issue (68)[https://github.com/tdwg/gbwg/issues/68]

[matentzn]

as long as it is valid yaml its fine

[matentzn]

In the workshop, we need to decide whether we should officially permit the use of IRIs in SSSOM or not. My sense is that parse should allow inputs with IRIs, but the canonical form should be in CURIE form. But I am happy to be convinced otherwise.