Add has_direct_output slot to Activity class

[pkalita-lbl]

Fixes #8

Summary

These changes bring the input and output slots on the Activity class more inline with the ShEx schema. In particular:

• Remove the existing has_direct_input slot
• Add has_input and has_output slots which are multivalued
• Add has_primary_input and has_primary_output slots which are single-valued

Details

• The range of all four of the new slots is MoleculeAssociation, which was previously used as the range of the has_direct_input slot.
• The MinervaWrapper.minerva_object_to_model method has been updated so that it populates the has_input, has_output, has_primary_input, and has_primary_output slots based on facts with the RO:0002233 (has input), RO:0002234 (has output), RO:0004009 (has primary input), and RO:0004008 (has primary output) properties, respectively.
• The MinervaWrapper.minerva_object_to_model method has been updated to emit warnings when the values of single-valued slots part_of, occurs_in, has_primary_input, or has_primary_output are overwritten. This is purely informative for now. The warnings may help us suss out areas where the modeling needs to be updated.
• A note about exclude_unset vs exclude_defaults in src/gocam/cli.py. I was seeing an issue where has_input and has_output were not appearing the YAML serialization produced by the gocam fetch command. The reason was that in the the LinkML-generated Pydantic model, multivalued slots have a default value of an empty list. The code in MinervaWrapper appended to those empty lists. As far as exclude_unset is concerned, mutating that default value means it is still "unset" (i.e. still equal by identity to the default). This didn't happen for the causal_associations slot because of a bit of code (now removed) that swapped out the default empty list for a new empty list. That's unnecessary, in general, but it did have the effect of satisfying exclude_unset. Since exclude_default does a compare-by-value with the default, it notices that the default value was modified in place and includes it in the output.

[github-actions]

PR Preview Action v1.4.7
🚀 Deployed preview to https://geneontology.github.io/gocam-py/pr-preview/pr-11/
on branch gh-pages at 2024-09-13 22:33 UTC

[pkalita-lbl]

After I opened this PR I did an exploration of running a lot more Minerva data through minerva_object_to_model and I'm seeing a fair number of these warnings about overwriting has_direct_input or has_direct_output. So empirically it would seem that maybe the choice of having those slots be single-valued need a second look.

[pkalita-lbl]

@cmungall I think this is now ready for review based on our discussion earlier this week.