cmungall
commented
5 years ago The use case and justification makes sense. Given we already house many application ontologies I see no reason not to include this.
Are you sure about data properties? So even if someone enters an ENVO class you would model as a string value with the ENVO label or ENVO CURIE? (UPDATE from https://github.com/GenomicsStandardsConsortium/mixs/issues/18 it seems like the string value may be a concatenation like <CURIE> [<LABEL>]. I think there is value in having the ability to make a 'normalized mixs' JSON-LD/RDF file where this is represented directly as a URI)
Do you see this as just housing properties, or also values? How do you intend to represent value sets (i.e. constraining properties to coming from a branch of an ontology, or from a defined set of values). Cc @mbrush who has thought a lot about this for clinical CDEs.
Although I generally don't like the strategy of distancing oneself from the science being modeled, I think in this case there may be value in modeling these things as information artefacts.
zhengj2007
ramonawalls
The Compliance and Interoperability Group (CIG) of the Genomics Standards Consortium would like to request an OBO Foundry namespace for metadata terms that are part of the Minimum Information for any (x) Sequence standard. We would like to use the prefix “MIXS”. Ramona Walls (rlwalls2008@gmail.com) will be the contact person, and our issue tracker is at https://github.com/GenomicsStandardsConsortium/mixs/issues.
MIxS terms are widely used to describe BioSamples and sequence data and are required by the INSDC databases. Despite wide adoption, the technical implementation of MIxS is out of date, and the CIG is working to modernize it. Permanent, redirectable, resolvable identifiers for terms are an important part of that effort.
Currently, most applications (including NCBI and EBI) simply use the term labels, either as the full label or as a linux-friendly short name. Biodiversity Information Standards (TDWG) is graciously providing hosting for resolvable URL names for MIxS (https://terms.tdwg.org/wiki/Class:MIxS), but these URLs are not redirectable. As we update our backend system, we would like to move to PURL-type identifiers whose corresponding landing pages we can control from within our own github organization.
We realize that MIxS terms do not make up an ontology, and therefore are outside the normal scope of the OBO Foundry. However, we think OBO PURLs would be a great home because 1) the infrastructure nicely suits our needs, 2) there is overlap between the OBO Foundry Operations Committee and the CIG (Lynn Schriml and Ramona Walls are members of both), which facilitates cross governance, 3) several OBO Foundry ontologies would like to reuse terms from MIxS, which will be much easier if they are in the OBO library, 4) repositories trying to implement the MIxS checklists (e.g., GigaDB, CyVerse) will find a stable list with auditable update tracking an invaluable tool.
We are open to suggestions about how to best organize the terms for inclusion within the OBO Library. My inclination is to create an OWL file that includes all of the MIxS terms as data properties. Then anyone who wants to use the MIxS properties in their ontology could just import them. Some of the properties, however, specify an ontology term as their target (e.g., https://terms.tdwg.org/wiki/mixs:env_feature), and those might be handled somewhat differently. It is a matter of sorting out how we want to handle MIxS as RDF, and that discussion is ongoing.