Add OAK matching to uPheno 2

obophenotype / upheno-dev

Framework for the automated construction of uPheno 2.0

MIT License

5 stars 5 forks source link

Add OAK matching to uPheno 2 #43

Open matentzn opened 1 year ago

matentzn commented 1 year ago

[x] Add OAK pipeline (including matching rules (same layout as mondo-ingest)
[x] matching rules must contain a rule to strip out all content between brackets
[x] For now, just download the dependency from https://bbop-ontologies.s3.amazonaws.com/upheno/current/upheno-release/all/upheno_all_with_relations.owl instead of trying to run the pipeline yourself.
[ ] YOu can create a subset of uPheno that is relevant like this:

robot filter -i upheno.owl --term http://purl.obolibrary.org/obo/UPHENO_0001001 --select "self descendants" -o upheno-minimal.owl

hrshdhgd commented 1 year ago

@matentzn , I may need a upheno.sssom.config.yml file for the mapping, correct?

matentzn commented 1 year ago

For which part exactly?

hrshdhgd commented 1 year ago

filtering prefixes. Do we need to do that? Similar to this.

matentzn commented 1 year ago

Yes, better.. There is sooo much noise we don't care about in uPheno!

hrshdhgd commented 1 year ago

Where do I grab that from? Or does that need to be curated?

matentzn commented 1 year ago

No great file atm, but you can search this file https://github.com/obophenotype/upheno-dev/blob/master/src/curation/upheno-config.yaml for prefix_iri to get a good starting point!

hrshdhgd commented 1 year ago

Ok, the robot code generates the upheno-minimal.owl file but after running oaklib, there aren't any mappings generated. So I used the entire file to generate the mappings and the results look promising. This takes around 22 mins.

cmungall commented 1 year ago

for this not to produce false positives like

DDPHENO:0000004 decreased migratory slug size (DDPHENO) skos:exactMatch DDPHENO:0000379 decreased vegetative cell size (DDPHENO) semapv:LexicalMatching oaklib 0.9411764705882353 oio:hasDbXref oio:hasDbXref pmid:18366659

we need the phenotype ontologies to follow standards and not use pmids as class xrefs

this may take a while to percolate, so should we filter at the robot level or the oak level?

hrshdhgd commented 1 year ago

I opened a parallel PR by accident. Sorry Nico!

matentzn commented 1 year ago

@cmungall i am happy to lead a push on cleaning out these xrefs, but first we need to answer this: https://github.com/information-artifact-ontology/ontology-metadata/issues/94