New Term Request: Minimal information standard

[delphinedauga]

Label

Recommendation: minimal information standard

Reasoning: a user requested to add the Minimum Information about Plant Phenotyping Experiment (MIAPPE) (https://fairsharing.org/bsg-s000543/) to the Minimum Information for Biological and Biomedical Investigations (MIBBI) collection (https://fairsharing.org/collection/MIBBI). But MIBBI is defined as a collection of the historical MIBBI foundry reporting guidelines (only the initial set of guidelines is listed). After discussion with Allyson and Pete, we decided to create a new term to address this issue.

IRI

Recommendation: require a new IRI.

Reasoning: Mass spectrometry ontology, Metabolomics Standards Initiative Ontology (MSIO) and nuclear magnetic resonance CV have « minimal information standard » as a child term of « standard » ; They have the same IRI. However, I found the definition of Wikipedia (which is the one we use for MIBBI) more appropriate. I would go with a new IRI.

Definition

Use Wikipedia’s: The minimum information standard is a set of guidelines for reporting data derived by relevant methods in biosciences. If followed, it ensures that the data can be easily verified, analysed and clearly interpreted by the wider scientific community. Keeping with these recommendations also facilitates the foundation of structuralized databases, public repositories and development of data analysis tools. The individual minimum information standards are brought by the communities of cross-disciplinary specialists focused on the problematic of the specific method used in experimental biology. 

Hierarchy

Recommendation: child of quality

Reasoning: We already have ‘FAIR’ as part of quality. I would go with the same level.

[allysonlister]

Thanks @delphinedauga - this is really thorough. I've assigned it to myself, but if anyone else wants the work please let me know! I will then probably send you a PR when I'm finished so that you can check my work.

As you have said, there are pluses and minuses to minting a new IRI for this term.

Benefits to a new IRI:

1. It will show that we are deliberately creating a concept distinct from the MS class that is used both by MS and MSIO
2. It will allow us to place it in the part of the hierarchy we deem most appropriate (Delphine has suggested as a quality and as a sibling of FAIR).
3. MS and MSIO place the class in different locations in their hierarchies. The originating ontology (MS) puts it only with a part_of relationship to "Proteomics Standards Initiative Mass Spectrometry Vocabularies", while MSIO puts it as a child of ICE -> standard. This inconsistency may not be what we want. We'd have to be careful where we pull the class from if we were to use the MS IRI so that we get the hierarchy we wish.
4. I agree that the definition that Delphine chose from Wikipedia is better than the existing one from the MS class.

Possible reasons to use the MS IRI:

1. I can see our new class being both a quality of something (it has the quality of being a minimal information standard), but I can equally see it represented as an information content entity. This is particularly relevant for us to consider if we wish to have the option to add more such terms in the future; in such a case, spending some time now to think about appropriate placement for standard subtypes might be be useful.
2. Would mean we don't need to maintain the class ourselves / mint an IRI.

Would be good to also hear from @Drosophilic on this. I would be happy to choose either option here, depending on which of these points is more important in this case.

Thanks!!

[allysonlister]

I've chatted with the OBO foundry community about this one. We noticed that it seems NMR:1000900 (MI standard) has a cross-reference to MS_1000899, which is "de facto standard" from MS, which is another child of "standard" and sibling to "MI standard". such a cross reference might be equivalent to something like skos:relatedMatch...

Nico from OBO Foundry says:

The central problem in application ontology development; and there is no great answer.

1. I would say first of all, application ontologies are the wild west. You do whatever you must to make it useful for your use case. So my comment here is: application ontologies should never be made available for re-use, just for use (for people with the same use case, say). This also means, do not submit application ontologies to OBO.
2. I think a “minting new” and mapping approach is often appropriate for application ontologies, but only of the ontology is used for something like search. If it is used for (meta)data curation, you will create a cost requiring the annotated data records to be actively linked back to the standard controlled vocabs they are mapped to.
3. “even though its reuse in other ontologies is in different places in the hierarchy?” -> you should never care what other ontologies say or do. The reference ontology that coined the term decides/defines it semantics. having other reference ontologies importing terms and then injecting axioms to change for example the position in the hierarchy is very bad practice

All of which I agree with. Philip Stroemert also reminds me that he believes NMRcv is intended as an AO as well, and therefore we probably should choose not to pull terms from there if we can help it.

We also already import MS and therefore would be a small change to continue importing them.

I am of the opinion that we should:

• use http://purl.obolibrary.org/obo/MS_1000900 as our minimum information standard
• place its parent term, standard, underneath ICE as is done with MSIO.
• disregard NMRcv as it is causing confusion and may be an AO itself.

This is what I'll progress with