Lessons learned

• SNPs constitute a fairly absurd cost in triples. Ensembl has 300 million rsIDs for human alone, which would quadrupal the size of their knowledgebase. This can not be managed right now.

• Adding/creating new concepts (text mining term lists, health benefit vocabulary, DSM example composite concept). Proprietary term lists made by experts in their field. Options discussed are:
  Use of existing terms or enhance thesaurus of EKP?
  Use of new terms (without reference to any ontology)
  a. Limit to use terms of official ontologies
  b. Add new terms (only when clearly defined, link to ontology not required for FAIRness)(not preferred by Euretos)
  c. Two step process: add new term and replace by official ontology term if it becomes available
  Submit new terms to ontology authorities (acceptance will take long)

• Have engineers involved in detailed workflow and data sources planning from the start

• Data preparation for ingestion/integration is laborious