Make a roundtrip test routine about gocamgen HOT 4 OPEN

pretty simple.. 2 commands to get from ttl tiles to gpad
./minerva-cli.sh --import-owl-models -j blazegraph.jnl -f ./test_gocam/
./minerva-cli.sh --lego-to-gpad-sparql --gpad-output ./gpad

All we really need now is a nice gpad diff and we are done.

from gocamgen.

Important to note that there are likely (certainly) going to be cases where there are differences in the output gpad that result from correct inferences during the export. Need to make the comparison ontology aware such that we can track when annotations are thus 'deepened'. These can be flagged for review and, assuming okay, should not count as errors or information loss when assessing conversion.

• investigate whether OWLTools is helpful here.. Otherwise will write a new semantic differ.

from gocamgen.

Send to @kltm for testing with gaf tools.

from gocamgen.

#29 provides a basic gpad diff procedure. We can test that no information is lost in the conversion by:

1. Running the conversion to GO-CAMs
2. Generating GPAD file for the GO-CAMs
3. Converting the GPAD file to a GAF file using OntoBio
4. Expanding the annotation set with all inferred annotations for this GAF file with Gaferencer
5. Converting the expanded GAF back to GPAD.
6. Running the gpad diff with source = initial MOD export GPAD and target = the expanded GO-CAM, GPAD from above.

That will tell us if anything is missing. If we want to investigate any new annotations generated from the conversion process and reasoning during gpad export, we can:

1. take the go-cam exported gpad as the source
2. run gaferencer to expand the initial MOD export gpad and set that as the target
3. run the gpad differ

gpad rows with no match are new.

from gocamgen.