search

BioSchemas / specifications

Issue tracker, technical wiki, and example markup
https://bioschemas.org
54 stars 52 forks source link

ID property should be mandatory for FormalParameters #543

Open albangaignard opened 3 years ago

albangaignard commented 3 years ago

This may lead to the introduction of RDF blank nodes in bioschemas markup. RDF blank nodes are "anonymous" nodes with a local ID without any unicity guaranty. This is a problem since (i) they may change each time the markup is generated, and (ii) they may collide when integrating multiple bioschemas markup in a single dataset or querying multiple bioschemas annotations.

AlasdairGray commented 3 years ago

For all markup we recommend the use of @id, @type, dct:conformsTo for every node in the graph; see issues #294, #297, #481.

The boilerplate code for adding this into the profiles is being worked on in branch standard-property-2; it got delayed by the website revamp, but should be merged in soon.

ljgarcia commented 3 years ago

We recommend but we cannot make it mandatory as there are cases where no id will be available. For instance, the protein has an id but the annotations, e.g., cellular location, will commonly not.

alaninmcr commented 3 years ago

The cellular location may have no sensible identifier field, but why should that prevent it having an @id.

I think the absence of @id will affect whether the data can be included in an RO-Crate.