apriltuesday
commented
1 year ago @M-casado Regarding the separate repo, I was thinking that it was providing an entirely different datasource to OT so would be more organised to separate it, especially given we're preparing to publish the ClinVar stuff. There will be some code reuse via libraries and executables but I think it's manageable, though it's a bit hard to predict as we haven't fully fleshed out the PGKB processing.
We can always move the code back to the eva-opentargets repo at a later date if that makes more sense.
Not even really a prototype as there are some fundamental missing features (see the many TODO's), but hopefully something to work off. In particular it doesn't explode on drugs/genes/phenotypes, and does not do any mapping to EFO/CHEBI identifiers or functional consequence prediction.
However it does provide a reproducible way to generate pseudo-evidence strings, which we can use to continue to develop the schema and requirements with Open Targets.
Test run here.