KG2c domain_range_exclusion property contains booleans in strings?

[amykglen]

I noticed today that in the kg2c_lite.json file, the domain_range_exclusion property on edges appears to be a python boolean in a string - e.g., "False" rather than false:

ubuntu@ip-172-31-52-1:~/plater-plover$ tail --bytes 1000 kg2c_lite_2.8.4.json 
     "domain_range_exclusion": "False",
      "object": "NCBITaxon:9606",
      "qualified_object_direction": "",
      "id": 45854429,
      "primary_knowledge_source": "infores:kegg",
      "subject": "KEGG:05416",
      "qualified_predicate": ""
    },
    {
      "qualified_object_aspect": "",
      "predicate": "biolink:subclass_of",
      "domain_range_exclusion": "False",
      "object": "NCBITaxon:9606",
      "qualified_object_direction": "",
      "id": 45854430,
      "primary_knowledge_source": "infores:kegg",
      "subject": "KEGG:05417",
      "qualified_predicate": ""
    },
    {
      "qualified_object_aspect": "",
      "predicate": "biolink:subclass_of",
      "domain_range_exclusion": "False",
      "object": "NCBITaxon:9606",
      "qualified_object_direction": "",
      "id": 45854431,
      "primary_knowledge_source": "infores:kegg",
      "subject": "KEGG:05418",
      "qualified_predicate": ""
    }
  ],
  "kg2_version": "2.8.4",
  "biolink_version": "3.1.2"
}

Would it make sense to store that value as an actual boolean, rather than a string?

(I'm not actually sure what this property is for - think I missed that over the summer - so I could be wrong, but just thought it's a bit odd while I was working with the kg2c_lite.json file today..)

[saramsey]

@sundareswarpullela do you know what JSON type that edge property is supposed to be?

[sundareswarpullela]

Its supposed to be a string as far as KG2c edge property is concerned. I consulted Erica about it when she implemented domain_range_exclusion for KG2pre. The current logic uses string comparison to filter the results from Plover at query time.

[sundareswarpullela]

There is no issue in changing the domain_range_exclusion key to a json boolean in kg2c_lite.json but this has to be done before triggering the KG2.8.6c build and there are some Expand changes necessary that filter domain_range_exclusion nodes. IMO this can be slated for Winter before the next KG2c version build.

[saramsey]

Thank you Sundar. Yes, we can include the change in the KG2.8.7c build.

[amykglen]

assigning @bazarkua as this would make a great first KG2c issue! (we discussed on a call this afternoon)

[bazarkua]

Updated create_kg2c_files.py and now domain_range_exclusion now property on edges is boolean rather than string.

tail --bytes 1000 kg2c_lite.json_TEST                                                                                                                                                    
xon:1040912",
      "primary_knowledge_source": "infores:ncbi-taxonomy",
      "qualified_predicate": "",
      "id": 281874,
      "domain_range_exclusion": false,
      "qualified_object_aspect": ""
    },
    {
      "object": "UMLS:C3356611",
      "predicate": "biolink:close_match",
      "qualified_object_direction": "",
      "subject": "NCBITaxon:1040913",
      "primary_knowledge_source": "infores:ncbi-taxonomy",
      "qualified_predicate": "",
      "id": 281875,
      "domain_range_exclusion": false,
      "qualified_object_aspect": ""
    },
    {
      "object": "UMLS:C3356612",
      "predicate": "biolink:close_match",
      "qualified_object_direction": "",
      "subject": "NCBITaxon:1040914",
      "primary_knowledge_source": "infores:ncbi-taxonomy",
      "qualified_predicate": "",
      "id": 281876,
      "domain_range_exclusion": false,
      "qualified_object_aspect": ""
    }
  ],
  "kg2_version": "2.10.2",
  "sub_version": "v1.0",
  "biolink_version": "4.2.5"
}```