Resources

All Resources have the following attributes:

• name: Human-readable name (required)
• description: Detailed description
• homepage_url: Primary resource URL
• repository: Version control repository
• license: Resource licensing information. When none is provided, the build fills this in from upstream sources and marks it status: inferred (see License Inheritance below)
• version: Resource version
• domain: Primary scientific domain (see Domains below)
• contacts: List of contact points
• products: Specific representations of the resource
• activity_status: Current state of the resource (active, inactive, etc.)
• use_instead: KG-Registry Resource ID(s) that replace this resource (see Pointing to a Replacement Resource below)
• tags: Categorization tags
• funding: Funding source information

Specialized Resource Types

Each Resource may have a more specific type.

KnowledgeGraph

A specialized Resource for graph-based data:

• Tracks graph creation and modification dates
• Lists contributors and components
• Represents nodes and edges data collection

DataSource

This Resource provides data. The data may or may not be in a graph structure.

DataModel

This Resource defines a data structure, i.e., a way to describe how data is organized and how its components relate to one another.

Ontology

This Resource defines an ontology, or a formal representation of a set of concepts within a domain and the relationships between those concepts.

Aggregator

This Resource combines multiple data sources.

License Inheritance

A KnowledgeGraph or Aggregator that provides no license is, by default, bound by the licenses of the sources it was built from. The build fills in license for such resources with the most restrictive license found among their sources and marks it with status: inferred. A license without a status is one the resource provided. The build never overwrites a provided license. To replace an inferred license, edit the license block and drop the status line and the inferred_from, unresolved_sources, and restrictiveness lines with it.

On the resource page and in the registry table an inferred license shows as the license label followed by “(inferred)”. Hovering over that word lists the sources that imposed it.

license:
  id: https://www.omim.org/help/agreement
  inferred_from:
  - omim
  - umls
  label: OMIM Use Agreement
  restrictiveness: custom
  status: inferred
  unresolved_sources: []

Sources are read from the resource’s components field and from the original_source and secondary_source associations on the products the resource owns. Only associations whose relation carries content count: prov:hadPrimarySource, prov:wasDerivedFrom, and prov:used. prov:wasInfluencedBy and prov:wasInformedBy are ignored.

Each source is resolved to a license in this order: the license on the named product, then the source resource’s own license, then the most restrictive license among that resource’s products, then the source resource’s own inferred license. A source with no license anywhere is listed under unresolved_sources and does not affect the result.

Licenses are ranked on a coarse ladder, least to most restrictive:

custom sits at the top because such terms have to be read before any reuse. A license is placed by its URL first and by its label when the URL is not recognized. A label that names several licenses lands on the most restrictive one it names.

The inferred license lists inferred_from, the sources at the winning tier, unresolved_sources, and restrictiveness, the tier itself. Inference runs during make and can also be run on its own:

make infer-licenses-dry-run
make infer-licenses
make license-report

The report (reports/license-inference.tsv) covers every KnowledgeGraph and Aggregator, including those with a declared license, and flags a conflict when the declared license is less restrictive than what the sources impose. A conflict is not an error in the registry. It is something to check with the resource’s maintainers.

Pointing to a Replacement Resource

Some resources are obsoleted or absorbed by others over time. When that happens we do not delete the older Resource’s page — its metadata still has historical and provenance value — but we point visitors toward whatever they should use in its place. This is done with the use_instead field, which holds one or more KG-Registry Resource IDs. On the Resource page these appear as a “Use instead” notice linking to the replacement(s).

We apply use_instead conservatively, only when both of the following hold:

• The resource is not active. Its activity_status is inactive, orphaned, or unresponsive. We do not redirect users away from an active resource, even if a similar one exists.
• There is a clear replacement. This is usually one of two situations:
  • A newer resource was explicitly intended to supersede the older one (for example, a version 2 released under a new identifier), or
  • The older resource has been subsumed by, merged into, or fully absorbed by another resource that now covers its content.

When neither a clear successor nor an absorbing resource exists — for instance a resource that was simply discontinued — we leave use_instead empty and rely on the activity_status banner alone. If a resource was split into several successors, or was subsumed by more than one resource, list each replacement ID.

The value of each use_instead entry must be an existing KG-Registry Resource ID (the id of the replacement’s page), not a name or URL.

Domains

Each Resource is annotated with one or more domains describing the kinds of information it provides. Wherever possible a domain maps to a concept in an external controlled vocabulary — the Medical Subject Headings (MeSH) or, where MeSH lacks a suitable concept, the NCI Thesaurus (NCIT) or EDAM. These mappings are recorded as the meaning of each value in DomainEnum and are shown as links on each Resource page.

Domains form a two-level hierarchy. The broad domains below describe a field. The specific domains that follow them name a narrower subject, such as a disease area or a kind of data, and each one records its broad domain with is_a. A Resource that lists a specific domain also lists its broad domain, so filtering by a broad domain still finds it. Use a specific domain when the Resource is substantially about that subject, not when the subject is one part of a larger aggregate.

Resources may cover one or more of the following domains:

Specific domains, grouped by the broad domain they belong to:

(The stub value is reserved for auto-generated placeholder pages and is not a subject-matter domain.)