i.e. extracting ontology class CURIes from env_broad_scale, env_loall_scale and env_medium

followup with Jasper, Peter, Paramvir and Mikaela

https://www.kbase.us/team/

• prototype in Jupyter if necessary, but create an application, possibly in another (NMDC or BBOP)? repo
• take an 80/20 low hanging fruit approach
• save intermediate data; make it easy to backtrack the results of the repairs
• keep track of possible losses "already added by me"
  • NCBI XML -> JSON/MongoDB (Jasper's or Mark's similar approaches) -> DuckDB
  • not all paths are added to DuckDB yet
  • attributes may be concatenated with ||| when writing into DuckDB ???
• new potential losses for benefit of performance/manageability
  • lowercase and normalize whitespaces?
• are there user-provided attributes that NCBI "failed" to harmonize
• take a stance on what the reasonable values are (see submission schema)
• suggest opportunities to route unreasonable values into other slots/fields/attributes?
• split wherever curies were found and make a unique collection of strings. annotate with multiple ontologies using OAK... using one combined backend or iteratively?

try to include drag remediation?

assume 0 or more occurrences of CURIes with {prefix}{delimiter}{local_id}

where prefix and local_id can be letters, numbers or both, with a min and max len

delimiters: mostly expect : or _. may include soem whitespace

there are some without delimiters... how many? ENVO1234567

just ENV instead of ENVO

ENVO:label pattern

NCIT etc with different length local_ids or with letters

some prefixes with numbers