identifiers.md: String representing a unique identifier

[korbinib]

If the identifiers specified in identifiers.md) are to be truly unique, this should be more explicit and the examples might have to be adapted as well, to include a 'truly' unique part.

[jahilton]

➕ to the above suggestion

Also, the definitions for several include "accession number", but the examples are not accessions. If accessions are the expectation, the definition could provide further information. For example, sample_id could be expected to be one of: a Sample accession from ENA (ERS..., DRS..., SRS...), a Sample accession from Biosample (SAM...), a Biosample accession from ENCODE (ENCBS...), etc.

[dbujold]

Thank you both for the comments. We've tried to clarify the context in which these ids are useful, this can now be seen in the pre_product_feedback branch. For this version of the checklist, we're focusing on "local" identifiers rather than accession numbers that are globally unique. We've now removed the mention of accession numbers.

[sveinugu]

For this version of the checklist, we're focusing on "local" identifiers rather than accession numbers that are globally unique. We've now removed the mention of accession numbers.

Could I ask for the reasoning behind this choice? Shouldn't globally unique (and persistent) identifiers be encouraged, in alignment with FAIR principle F1: (Meta) data are assigned globally unique and persistent identifiers?

[sveinugu]

Suggestion: In the adoption by the RDA working group: "FAIRification of Genomic Annotations" we have added external variants of all the identifiers, e.g. experiment_external_id (row 51) and sample_external_id (row 158). Perhaps this could be included in the standard itself?

[sveinugu]

#8 and #11 also comment on internal vs external identifiers

[Woolly-at-EBI]

I think that collectively we are not far apart. Yes the findability/discoverability is what the main focus has been.
In the project we are still just earlier in the maturity, agree that in implementation the identifiers nomenclature need to be nailed down.
Indeed Sveinung, in INSDC there are a mix of a local and global identifiers. (that could be a lot better than it currently is!).

Which of URIs or CURIE is the best format for identifiers seems to cause some warfare, personally whatever is fine as long as the identifier can be validated.

I think adopting much of the practices that are in the RDA sheet is a good way to go and saves re-inventing wheels, + meaning increased interoperability across more to the scientific community. (completely out of the scope for the sequence_experiments, I noticed the species_taxon and sex examples, not initially a fan of those examples, I obviously will need to know more.)

[dbujold]

As for internal VS external ids, I go with Peter to say that local IDs is really a first step. And that archival sites IDs considerations could be the topic of a future revision. This doesn't disqualify what will have been added in v1, both are complementary.

As for CURIE vs URIs, same as Peter, I don't have an opinion. If RDA already recommends one way vs the other, I'm fine with aligning with that.