Molecule dataset for release 24.06 has multiple Unknown values: inconsistent casing

[dhimmel]

@related-sciences appreciates all the great work by the OT team and noticed something small when upgrading to 24.06.

Running the following on BigQuery, which currently is based on the 24.06 release:

SELECT
  drugType,
  COUNT(drugType) AS drugTypeCount,
FROM
  `open-targets-prod.platform.molecule`
GROUP BY drugType
ORDER BY drugType

Produces the following table of drugType counts:

drugType	drugTypeCount
Antibody	963
Antibody drug conjugate	119
Cell	52
Enzyme	91
Gene	117
Oligonucleotide	159
Oligosaccharide	52
Protein	741
Small molecule	14854
Unknown	870
unknown	23

Notice the mixed casing for the "Unknown" / "unknown" value. This issue also exists in 24.03 although prior releases have been consistent in only using "Unknown".

Versioned GCS path gs://open-targets-data-releases/24.06/output/etl/parquet/molecule.

There's the narrow fix and then possibly a broader fix of selecting possible values from an enum or applying a schema that would prevent an issue like this from ever occurring.

[prashantuniyal02]

Hi @dhimmel, this issue has been resolved. The changes will be released in the upcoming 24.09 platform release with the following drugType:

drugType
Antibody
Antibody drug conjugate
Cell
Enzyme
Gene
Oligonucleotide
Oligosaccharide
Protein
Small molecule
Unknown