Discussion: Using "instance of" to indicate the semantic type

[stuppie]

This is a record of previous discussions with Wikidata community members regarding

• using "instance of" to indicate the semantic type
• classes being instance of one of their superclasses

https://www.wikidata.org/wiki/Wikidata:Project_chat#ProteinBoxBot_and_biological_processes

https://www.wikidata.org/wiki/User_talk:ProteinBoxBot/Archive_2#subclass_relationships

https://www.wikidata.org/wiki/Wikidata:Project_chat/Archive/2016/04#Instance_of

https://www.wikidata.org/wiki/Wikidata:Requests_for_comment/Are_colors_instance-of_or_subclass-of_color

https://www.wikidata.org/wiki/User_talk:ProteinBoxBot#To_prevent_items_from_being_P31_and_P279*_of_the_same_class

[andrawaag]

https://www.wikidata.org/wiki/User_talk:ProteinBoxBot/Archive_2#Instance_of_disease

[stuppie]

How do we handle "special" genes? Issues: A psuedogene is not a gene. Genes and transcripts are conflated onto one item. Pseudogene: https://www.wikidata.org/wiki/Q17709268 ncRNA: https://www.wikidata.org/wiki/Q18048938 Protein-coding gene: https://www.wikidata.org/wiki/Q21172174

Option A: All protein-coding genes are instance of gene, and messenger RNA pseudogenes are instance of pseudogene ncRNA: instance of gene and ncRNA What to do with ncRNA ? Instance of ncRNA and instance of gene? Issue: We currently conflate gene and RNA items. (genes have gene and transcript IDs on them. proteins are a separate item)

Option B: Create new property: Type of gene: {'protein-coding', 'ncRNA', 'pseudogene' … } https://www.ncbi.nlm.nih.gov/gene?cmd=retrieve&dopt=default&list_uids=1017 https://www.ncbi.nlm.nih.gov/gene?cmd=retrieve&dopt=default&list_uids=114036

Option C: Align with the bioschemas approach as seeing it as a record

Option D: Split into: gene, transcript, protein. Align with SO

For now: Leave both subclass and instance of..