Open sofyalaski opened 2 months ago
In what field? Can you give an example?
Me and Yves found no such case in fact. In theory, if we have an array of authors and if one author has many addresses it would be an array of arrays, which is not currently supported. Although if we look at OneDep, they also only require one address in fields we propagate
Sounds like that would be an array of objects, where the objects have array fields. Is that supported?
I know mmCIF does use matrices for some fields (eg pdbx_struct_oper_list, used in crystallography). Would this be unsupported?
To your first question: not for now, as we had no real use case. If there's such a case in OSCEM, where we have a value, which is an array of objects, this would translate to a loop_
-kind of structure.
As you can see in your example, the allowed Category Data Items have a predefined matrix dimensions. I think that is something else.
If we really have a case, where our value is an array of objects, and one of objects values is an array, this would need to translate like that for that authors example:
First, json example ( I added a second email here to Yves' data as example)
{
"authors": [
{
"name": "Tittes",
"first_name": "Yves",
"email": [
"yves.tittes@unibas.ch",
"second@email.com"
],
"orcid": "0000-0002-7026-7646",
"name_org": "University of Basel",
"type_org": "Academic",
"country": "Switzerland"
},
{
"name": "Rodriguez",
"first_name": "Carlos F.",
"email": "cf.rodriguez@unibas.ch",
"orcid": "0000-0001-9166-0132",
"name_org": "University of Basel",
"type_org": "Academic",
"country": "Switzerland"
}
]
}
this will look like that in PDBx:
_pdbx_contact_author.email
_pdbx_contact_author.name_first
_pdbx_contact_author.name_last
_pdbx_contact_author.country
_pdbx_contact_author.phone
_pdbx_contact_author.role
_pdbx_contact_author.organization_type
_pdbx_contact_author.identifier_ORCID
yves.tittes@unibas.ch Yves Tittes Switzerland '+41 61 207 74 57' 'responsible scientist' academic 0000-0002-7026-7646
second@email.com Yves Tittes Switzerland '+41 61 207 74 57' 'responsible scientist' academic 0000-0002-7026-7646
cf.rodriguez@unibas.ch 'Carlos F.' Rodriguez Switzerland '+41 61 207 68 22' investigator academic 0000-0001-9166-0132
As you can see, Yves has two entries then. The other thing is, that this propagates to OneDep as three authors, which is not correct either I think? But this is the only example of which Yves and I could think so far. Nothing really applicable to sample or instrument. Maybe we discuss it again once Scipion schema is ready. In case they have such a case I will add the support
Description
If json file contains array of arrays, these will not produce a valid mmCIF as of now.
Are there any use cases where it might be neccessary?