DATASET_LIST.yml

# The master list for the translator Datasets,
# which are subject to convert to an API.
#
# Some basic metadata fields are defined as:
#   "name": a short name for the dataset
#   "description": a longer description about the dataset
#   "url": a url that points to the dataset
#   "translator"  translator-specific attributes:
#     "enpoint"  dataset access information
#       "url"  endpoint url
#       "biolink_model"  yes/no - whether the dataset conforms to the biolink data model
#       "source_format"  a format of the source file/API
#       "api_type"  a type of the API (OpenAPI/smartAPI/REST/knowledge-beacon/...)
#       "api_registry"  a link to a registry contaninig additional information about the endpoint
#       "parse_code"  a link to code used to parse/generate the dataset
#     "categories"  categories of data contained in the dataset


datasets:
  - name: BioLink
    description: Monarch BioLink API
    url: https://monarchinitiative.org/page/services
    translator:
        endpoint:
          - url: https://biolink-kb.ncats.io/beacon/biolink/
            biolink_model: yes
            source_format: API/JSON
            api_type: knowledge-beacon
            api_registry: 
            parse_code: 
          - url: https://api.monarchinitiative.org/api/
            biolink_model: yes
            source_format: API/JSON
            api_type: knowledge-beacon
            api_registry: https://github.com/NCATS-Tangerine/translator-api-registry/blob/master/biolink/openapi.yml
            parse_code:
        categories:
          - sequence feature 
          - gene 
          - anatomical entity 
          - phenotype 
          - cellular component 
          - cell 
          - disease 
          - quality 
          - variant locus 
          - sequence alteration 
          - molecular entity 
          - clinical_course 
          - snv 
          - organism 
          - genotype 
          - inheritance 

  - name: HMDB
    description: Human Metabolome Database
    url: http://www.hmdb.ca/
    translator:
        endpoint:
          - url: http://www.hmdb.ca/system/downloads/current/hmdb_metabolites.zip
            biolink_model: no
            source_format: XML
          - url: https://translator.ncats.io/hmdb-knowledge-beacon
            biolink_model: yes
            source_format: API/JSON
            api_type: knowledge-beacon
            api_registry: 
            parse-code: https://github.com/NCATS-Tangerine/HMDB-knowledge-beacon
        categories:
          - metabolite 
          - disease 
          - pathway 
          - cellular location 
          - biospecimen 
          - tissue 
          - protein 

  - name: PubChem
    description: PubChem is an open chemistry database at the National Institutes of Health (NIH).
    url: https://pubchem.ncbi.nlm.nih.gov/
    translator:
        endpoint:
          - url: https://pubchemdocs.ncbi.nlm.nih.gov/power-user-gateway
            biolink_model: no
            source_format: API/XML
            api_type: custom
        categories:
          - chemical substance
          - assay