Boomer pipeline prototype implementation for Mondo ICD10CM mapping

[matentzn]

Boomer pipeline prototype implementation for Mondo ICD10CM mapping

[matentzn]

@cmungall can you just roughly jot down the method, e.g. something like:

1. collect all mappings
2. (unify to skos? or can the input be mix of skos and owl)
3. generate individual ptables for each input table
   • (how to provide global priors on mapping set level?)
   • Or: merge everything first into one massive SSSOM file?
4. run boomer
5. Look at X (what is X) make a fix, run again

Thanks.

[cmungall]

1. Generate OWL for each source
   • can we flat
   • should use subClassOf where appropriate (I think this only applies to ORDO - transform the part-ofs to is-a)
2. Aggregate all mappings as SSSOM files
3. Select subset of ontologies mappings for any given experiment (this may be simply everything to start with)
4. Merge selected OWL ontologies using robot
5. Fill a default confidence value for each SSSOM file
   • we can iterate on more sophisticated strategies, just choose 0.8 as a default for now
6. Extract a ptable from all selected SSSOM files.
   • just use the default sssom-py method for now, no parameter tweaking
   • eventually boomer will take sssom natively INCATools/boomer#47
   • I don't think sssompy will make the prefixes table for boomer but this would be good functionality to have
7. Feed the ptable and the merged ontology to boomer
   • use defaults for now
8. Explore output

• working on documenting this: INCATools/boomer#261
• but for now if you see a folder full of json and a md file, consider it a success

The input to boomer should be a ptable and the combined OWL files of all ontologies. We can iterate on methods to set priors but just to get started, something like 0.85/0.5/0.05/0.5.

[matentzn]

Something to look at: https://github.com/geneontology/go-rhea-boom/blob/master/Makefile

[matentzn]

This ticket is now a bit too general, I consider this done - we will replace this now moving forward by more specific issues