Implement dataset properties UI

Dataset and Collection metadata will be helpful context for users to understand the data and additional information about the collection. Please include the following config fields in the data overview drawer on both desktop and hosted:

Title

Dataset Metadata

• organism, organism_ontology_term_id
• assay, assay_ontology_term_id
• tissue, tissue_ontology_term_id
• cell type, cell_type_ontology_term_id
• disease, disease_ontology_term_id
• ethnicity, ethnicity_ontology_term_id
• development_stage, development_stage_ontology_term_id
• sex
• layer_description

Collection Metadata

• project_name (Collection Name)
• project_description (Collection Description)
• publication_doi or preprint_doi
• contributors
  • Name
  • Institution
  • Email
• project_links
  • Link Name
  • Link URL
  • Link Type

If any fields are not populated, please exclude. Out of scope: verifying publication information by hitting the crossref API.

Old Notes: Most likely this would be data sourced from H5AD .uns or Loom /attrs/. We will need some backend convention for specifying which are to be displayed and in what way.

Implementation: Where’s the data found to fill this drawer?

• corpora-config-props
• properties moving over from the left sidebar.
• But to get journal name AFTER publication, might need to hit crossref API to get correct information. (backend)
• will need a test in cxg to ensure we don’t break data existing in config-props
• only expected to work cleanly for remix datasets, should work for all data sets.
• move singletons out into sidebar all the time (not just corpora)

[seve]

@tihaun Since this ticket is migrating over to implementation work I think it’s best to tweak the estimate to a 4.

@colinmegill How does that align with what you have in mind?

[mckinsel]

• We have the required ontology ids/labels: organism, assay, tissue, disease, ethnicity, development_stage, sex. You could put those here and pull them out of the LSB when they’re singlets.
• There are more links. Right now cellxgene only displays the “summary” link as the about link in the upper left, but datasets usually have more: more ways to access raw data, a link to the lab website, a github repo with some of their analysis code, etc.
• In the mock, authors are presented as an attribute of the dataset. But, I think the right model is datasets may have an associated preprint or publication, and authors are an attribute of that. There’s a little more info you might want to drop in from the crossref API them: title, journal info, publication date
• Distinct from authors, we have a notion of “contributor”. This is the person(s) who actually gave us the data and is the best contact for data questions. We usually get their name, email, institution.
• Datasets are members of collections (née projects), which have a name and description, so maybe like “A part of …”