jamesamcl

We have the test framework in place now, so next we just need to compare `testcases_expected_output_api` in this repository with actual output from OLS3.

Probably easiest if I do this offline and then open issues for any disparities. Then once `testcases_expected_output_api` matches OLS3 we have correct integration tests forever.

JSON Lines also works very well with Pyspark.

Yes you can swap “properties” in the URL for “terms”

Looks like UniProt also acts as a registry definining prefixes for lots of other DBs, e.g.: ``` ``` (taken from [one of the RDF files in the latest release](https://ftp.ebi.ac.uk/pub/databases/uniprot/current_release/rdf/uniprotkb_reviewed_eukaryota_opisthokonta_metazoa_33208_0.rdf.xz)) Not...

Full list of databases obtained from sparql endpoint, though I am not sure exactly how these map to the URI prefixes seen above: ``` http://purl.uniprot.org/database/EnsemblFungi http://purl.uniprot.org/database/EnsemblProtists http://purl.uniprot.org/database/EnsemblMetazoa http://purl.uniprot.org/database/PDB http://purl.uniprot.org/database/Ensembl http://purl.uniprot.org/database/EnsemblPlants...

@JervenBolleman Apologies for tagging you directly if this was inappropriate; I was under the impression you were responsible for the RDF serialization of UniProt and our culture has generally been...

They still look wrong to me? "short_form" : "CAO_CAO_000001", "obo_id" : "CAO_CAO:000001",

Possibly related: https://github.com/EBISPOT/ols4/issues/721 ?

> This is amazing, a great leap forward. > I think everyone agrees that one should be using their package manager for most of their software. > But especially for...