Dataset-JSON 1.1 - Githubissues

TeMeta commented 6 months ago

Dataset-JSON 1.1 major version

Overhauled schema based on community feedback
Switched from TS to LinkML for more varied modelling support
Added tooling for conversion from 1.0 to 1.1
Added Jupyter Notebook for easier interaction
Updated all examples

jehugaleahsa commented 5 months ago

Should the dataset-json version be updated to 1.1.0 in these files?

TeMeta commented 5 months ago

Need to set datasetJSONVersion to "1.1.0"

lexjansen commented 5 months ago

I think it would be good to add the SEND examples from https://github.com/cdisc-org/DataExchange-DatasetJson/pull/21 as well. We want to make sure SEND is represented.

lexjansen commented 5 months ago

The examples need some refinement. Example adam\adlbcpv.json:

        {
            "dataType": "integer",
            "displayFormat": "DATE9.",
            "itemOID": "IT.ADLBCPV.TRTSDT",
            "keySequence": 0,
            "label": "Date of First Exposure to Treatment",
            "length": 8,
            "name": "TRTSDT",
            "targetDataType": "integer"
        },

In this case dataType should be "date". Also. the keySequence should not be 0 when the item is not part of key variable in the dataset structure. I would prefer not to have the keySequence key then.

Also, not wrong, but:

if targetDataType is the same as dataType, do we need to specify it?
if displayFormat is null, do we even need to have the key.
I would prefer the JSON files not be pretty printed, so that people get used to this. You really don't want pretty-printing in a submission. Common tools can format with indentaton on-the-fly.

TeMeta commented 5 months ago

I think it would be good to add the SEND examples from #21 as well. We want to make sure SEND is represented.

Good point. Can we do it after this PR has been corrected and approved, then update the SEND examples the same way upon merging #21 to the version 1.1?

lexjansen commented 5 months ago

Looks good! One more use case to illustrate targetDataType. In the SDTM datasets there are decimal data types. They also need targetDataType.

Example:

        {
            "dataType": "decimal",
            "itemOID": "IT.VS.VSSTRESN",
            "label": "Numeric Result/Finding in Standard Units",
            "length": 8,
            "name": "VSSTRESN"
        },

I think this would be:

        {
            "dataType": "string",
            "itemOID": "IT.VS.VSSTRESN",
            "label": "Numeric Result/Finding in Standard Units",
            "length": 8,
            "name": "VSSTRESN"
            "targetDataType": "decimal",
        },

lexjansen commented 5 months ago

Great that you added the order! I think you missed a few:

    "metaDataRef": "https://metadata.location.org/CDISCPILOT01/define.xml",
    "metaDataVersionOID": "MDV.MSGv2.0.SDTMIG.3.3.SDTM.1.7",
    "studyOID": "cdisc.com/CDISCPILOT01",
    "itemGroupOID": "IG.VS"

TeMeta commented 5 months ago

Great that you added the order! I think you missed a few:

    "metaDataRef": "https://metadata.location.org/CDISCPILOT01/define.xml",
    "metaDataVersionOID": "MDV.MSGv2.0.SDTMIG.3.3.SDTM.1.7",
    "studyOID": "cdisc.com/CDISCPILOT01",
    "itemGroupOID": "IG.VS"

Hi Lex, this has been fixed now. I used vs.json from 1.0 as the reference for ordering elements Thanks for the thorough review!

lexjansen commented 5 months ago

@swhume It is:

        {
            "itemOID": "IT.VS.VSSTRESN",
            "name": "VSSTRESN",
            "label": "Numeric Result/Finding in Standard Units",
            "dataType": "string",
            "targetDataType": "decimal",
            "length": 8
        },

cdisc-org / DataExchange-DatasetJson

Dataset-JSON 1.1 #23