Dataset object populated with properties from h5ad uns

[ubyndr]

Fixes #67

Example output;

ns:9698a8ea-19de-4273-acea-3728a4180fc1 a obo:PCL_0010001 ;
    rdfs:label "Splatter" ;
    ns:supercluster_term "Splatter" ;
    obo:RO_0015003 ns:d9f4490f-9801-4835-b52a-60e2423d259f ;
    dcterms:source ns:c1ca0cd3-144e-469c-84ae-cc0081e1f6a1 .

ns:d9f4490f-9801-4835-b52a-60e2423d259f a [ a owl:Restriction ;
            owl:onProperty obo:RO_0002473 ;
            owl:someValuesFrom obo:CL_0000540 ],
        obo:PCL_0010001 ;
    rdfs:label "neuron" ;
    ns:cell_type "neuron" ;
    dcterms:source ns:c1ca0cd3-144e-469c-84ae-cc0081e1f6a1 .

ns:c1ca0cd3-144e-469c-84ae-cc0081e1f6a1 a schema:Dataset ;
    rdfs:label "dataset" ;
    ns:citation "Publication: https://doi.org/10.1126/science.add7046 Dataset Version: https://datasets.cellxgene.cziscience.com/0bb62fec-0cf1-46e1-9d10-de65a6d4f814.h5ad curated and distributed by CZ CELLxGENE Discover in Collection: https://cellxgene.cziscience.com/collections/283d65eb-dd53-496d-adb7-7570c7caa443" ;
    ns:obs_meta "[[{\"field_name\": \"ROIGroup\", \"field_type\": \"author_cell_type_label\"}], [{\"field_name\": \"ROIGroupCoarse\", \"field_type\": \"author_cell_type_label\"}], [{\"field_name\": \"ROIGroupFine\", \"field_type\": \"author_cell_type_label\"}], [{\"field_name\": \"cluster_id\", \"field_type\": \"author_cell_type_label\"}], [{\"field_name\": \"dissection\", \"field_type\": \"author_cell_type_label\"}], [{\"field_name\": \"roi\", \"field_type\": \"author_cell_type_label\"}], [{\"field_name\": \"sample_id\", \"field_type\": \"author_cell_type_label\"}], [{\"field_name\": \"subcluster_id\", \"field_type\": \"author_cell_type_label\"}], [{\"field_name\": \"supercluster_term\", \"field_type\": \"author_cell_type_label\"}]]" ;
    ns:schema_reference "https://github.com/chanzuckerberg/single-cell-curation/blob/main/schema/5.0.0/schema.md" ;
    ns:schema_version "5.0.0" ;
    ns:title "All neurons" .

I am currently using the following structure for the Dataset class:

DATASET = {"iri": "https://schema.org/Dataset", "label": "dataset"}

I am open to modifying this as per your preferences. Please let me know if you would like any changes.

[dosumis]

Looks good. Thinking about it, the main use of the CxG citation field is probably to provide a blob of text that authors can paste into their publications. For Indexing purposes I think important to have contents split into distinct fields too.

dc:publication: DOI
Dataset_version: ... # Is there a schema.org property for this?
Collection: ...

Could be additional ticket as need to think through and co-ord how this will work with SCXA h5ad.