Odd objects in ZP has_part edges

[kevinschaper]

I'm not sure what to make of objects in these associations

id      subject predicate       object  category        relation        knowledge_source
urn:uuid:72275123-b615-40e5-a18e-d8c6270456aa   ZP:00000000     biolink:has_part        N8e0099d3a3d94250b35165a29c5d231f       biolink:Association     BFO:0000051     Graph
urn:uuid:158d58ec-a54a-43a9-a2a8-396258cf7af2   ZP:0000002      biolink:has_part        Ne5c13f76e6274f479508b2bc485d4b93       biolink:Association     BFO:0000051     Graph
urn:uuid:4760c510-0ef1-47a6-ac5a-d2965f61dd15   ZP:0001021      biolink:has_part        N3a610fb4ab51494dbde6cd7a4344f2a3       biolink:Association     BFO:0000051     Graph
urn:uuid:3554a3dd-a769-4709-b6db-814e69e2798a   ZP:0000605      biolink:has_part        N75cfe3bd80534e71b6ede82ebe499040       biolink:Association     BFO:0000051     Graph
urn:uuid:894d2355-dd06-4663-b9f7-65f51653e0f5   ZP:0005505      biolink:has_part        N367cc09804b344d982385355f5888ab0       biolink:Association     BFO:0000051     Graph
urn:uuid:802d1217-3b82-4a6e-8b2a-e4762c452723   ZP:0010064      biolink:has_part        N2d14d99b853a4d0ea6aee778b76f1a04       biolink:Association     BFO:0000051     Graph
urn:uuid:f5188cdb-1ed5-4528-b299-ac58ce2aa3bf   ZP:0011389      biolink:has_part        Nec32d573a92c4adaa87f7896dc7fdefc       biolink:Association     BFO:0000051     Graph
urn:uuid:64be5935-e981-4ba0-8c8d-2994c4f732ca   ZP:0019005      biolink:has_part        N5fb7aeb0c69241d8be016d9b10c92c63       biolink:Association     BFO:0000051     Graph
urn:uuid:bf5eeca5-5b7b-480e-922a-4b0fea0a62fd   ZP:0014862      biolink:has_part        Nd3daf5295b5d4335a16047d63a65cd8c       biolink:Association     BFO:0000051     Graph
urn:uuid:bda73968-0c10-415c-b9f9-280ed0e4a539   ZP:0004433      biolink:has_part        N1e2d4df8a6a045b49d2f48ae92b2003f       biolink:Association     BFO:0000051     Graph

finding one in the nodes file, it's just:

N8e0099d3a3d94250b35165a29c5d231f       biolink:NamedThing                              Graph   

[caufieldjh]

Those are blank nodes - they really shouldn't be there, but here they are, and they'll be gone in subsequent releases.

[caufieldjh]

That being said, it's very good to know where they're still appearing, because it's revealing where the subq reconstruction may not be working as expected (or some other parsing issue is happening)

[matentzn]

Which part of the process where you thinking for stripping anonymous (blank) nodes? phenio or KG-Phenio? I would tent a bit to the latter because there are use cases for phenio that could benefit from the logical axioms..

[caufieldjh]

The bnodes really shouldn't be in KG-Phenio at all so at minimum the graph transform step can remove them. KGX will catch and ignore bnodes when they're created during a transform process (i.e., changing serialization formats) but not when they're created upstream, as is happening here during subq reconstruction. I expect some of these will disappear once we've made that part (subq reconstruction) more comprehensive.