ISA Context Problems

[floWetzels]

ISA json-context files from the isa-api have the following semantical problems for the conversion to RO-Crate

Mapping leading to problems with RO-Crate

• Contexts overall do not fit to RO-Crate conventions, since all studies and assays (subfolders in an ARC) should be sdo:Dataset:

  Data Entities representing directories MUST be of "@type": "Dataset"

  see RO-Crate specification.

Mapping leading to illegal relationships

• isa:Assay mapped to sdo:MedicalTest, isa:measurementType mapped to sdo:variableMeasured, but the latter is not part of sdo:MedicalTest, even though such a relationship exists in the datamodel.
• isa:characteristicCategories mapped to sdo:measuredProperty, but the latter not part of sdo:MedicalTest.
• isa:Study mapped to sdo:MedicalStudy, isa:studyDesignDescriptors mapped to sdo:trialDesign, but the latter is not part of sdo:MedicalStudy.
• isa:affiliations of isa:Person mapped to sdo:affiliation of sdo:Person, but types don't fit (Text vs Organization). Where does isa:Organization come from? Not part of the specification.

Mapping of non-existent fields

• Assay context maps identifier, name, description, but they are not properties of isa:Assay.

Mapping which doesn't fit semantically

• isa:Sample to sdo:Product
• isa:Protocol to sdo:HowTo
• isa:Comment to sdo:Comment (they describe different things!).
• isa:material mapped to sdo:participant.

Thus, I suggest to write ARC/RO-Crate specific context files manually.

[HLWeil]

I think we can close this, having created our own profile in https://github.com/nfdi4plants/isa-ro-crate-profile? @floWetzels

[floWetzels]

This is no longer relevant for us, so I think you're right. But I guess this is still something to keep in mind to talk to some of the ISA people to? Not sure...