Further define bioinformatics taxonomy facets

[khawkins98]

Currently we have a high level of the IA facets at https://embl-design-language.github.io/Springboard/information-architecture/#facet-structure-and-categories

They are:

1. Who
   - people
   - groups and teams
1. What
   - research
      - Cell Biology and Biophysics
      - Developmental Biology
      - Genome Biology
      - Structural and Computational Biology
      - Directors' Research
      - Tissue biology and disease modelling
      - Mouse biology
      - Bioinformatics (EBI)
   - services
      - online services (EBI)
      - core facilities
   - training
      - online/external
        - online bioinformatic services (EBI)
      - collaboration
   - industry transfer
   - inspiration
   - about
     - news
     - jobs
     - mission statement
     - organisational information
1. Where
   - Heidelberg
   - Barcelona
   - Grenoble
   - Hamburg
   - Hinxton
   - Rome

Not expressed are the bioinformatics sub categories:

• DNA & RNA
• Gene Expression
• Proteins
• Structures
• Systems
• Chemical biology
• Ontologies
• Literature
• Cross domain
• Type
  • Database
  • Tool

These would apply under:

• what:services:online-services
• what:research:bioinformatics
• what:training:training:online:bioinformatics

It would be redundant to list the bioinformatics sub categories three times, so instead it should be listed as a "special meta-taxonomy"(?), along with EDAM, I think.

[khawkins98]

After a chat with @jonathanhickford , I think this bioinformatic taxonomy is worth having, but not part of our "Central EMBL IA". Rather a supplemental one that can be used by content and pages, as can EDAM.

[jonathanhickford]

Potentially silly question. Can content live under multiple values for the what taxonomy? E.g. can I be bioinformatics and mouse biology research?

[khawkins98]

I'd think the answer is that content should live (be presented) anywhere a user expects to find it.

So the answer, I'd say: quite possibly, yes, content may be presented in multiple locations. But we definitely don't want to maintain it in two locations, so we need a system that allow content to be queryable.

[markboulton]

Let me see if I understand this. Are we saying that content has a single source, but can be reached through several routes and have different taxonomy labels?

[khawkins98]

Yes/no 🙃

Content is accesable through many paths, but has one canonical URL.

Here's how I'm imagining it:

[jonathanhickford]

That right.

Say it's biographical content about Paul Flicek (who for context runs multiple EMBL-EBI bioinformatics services, and a research team that research mouse biology and genome biology) he might end up something like:

• Who = person:paul_flicek, team:paul_flicek_research, team:ensembl_service
• What = research:mouse_biology, research:gemome_biology, service:online:enesmbl, service:online:ipmc_portal, service:online:IGSR
• Where = Hinxton

On that page we might want to:

• list jobs in hinxton in Pauls teams
• link to the services he is involved in
• link to other research groups who do genome biology or mouse biology

[khawkins98]

I think @jonathanhickford and I may have posted at the same time, but I think we're saying the same thing, or at least the same thing from slightly different angles.

[jonathanhickford]

Tag driven site war story one:

It was an educational site about animals, and was dynamically tag and taxonomy driven. There were lots of photos of animals, some with individual animals (tag: elephant) others with many animals in (tag: elephant, tag:giraffe).

The issue was that as on the page about elephants it would quite often end up with a featured image lablelled 'elephant' that was of a giraffe, with a tiny elephant somewhere far off in the background.

That ended up with a fairly complex system of 'primary' tags and 'secondary' tags - which was miserable and the project was eventually cancelled.

We perhaps mitigate against that here by having editorially controlled 'content' pages. However I think there's a risk we end up with the problem occuring at the 'directory' or 'facet' level. For example does every page on the website seem to offer you links to Paul Flicek.

This is probably an issue for much further down the line, and I'm perhaps jumping the gun in thinking that the facets are not manually curated, which we could always do for the more complex ones.

[markboulton]

Ok, got it. Thanks for the clarification, both.