Reduce query ids by only using bound nodes for pfocr figure search

[tokebe]

Doesn't address all relevancy problems due to creative mode interactions (pfocr augmentation happens at template-level, therefor not grabbing explicitly the bound nodes in the final inferred response), however this should ensure slightly higher relevancy of a given figure to the template it came from, in turn making it slightly more relevant to the overall inferred result.

Most importantly, because only the bound nodes are used, fewer ids are being used to query pfocr, which should decrease overall pfocr search time. Full traverseResultforNodes is still used in scoring and for matchedQueries, so the score should still be slightly higher-quality.

[colleenXu]

Related to previous work here biothings/biothings_explorer#837, probably trying to improve the amount of time spent on pfocr result-augment queries?

Also maybe related to biothings/biothings_explorer#847 (comment)

[colleenXu]

@tokebe

I see something odd in pfocr-test-creative-increases-ADRB2.json.zip.
I found some matched curies in the first few results' pfocr sections that aren't in the KG at all. I suspect that they come from the result nodes' IDs from other namespaces (mistakenly treated as NCBIGene IDs)

• NCBIGene:285 (ANGPT2 in the linked figure)
  • VS the result does have the intermediate ADRB1 (NCBIGene:153), which has the equivalent ID HGNC:285
• NCBIGene:5816 (PVALB in the linked figure)
  • VS the result does have the chem (-)-adrenaline ("CHEBI:28918"), which has the equivalent ID PUBCHEM.COMPOUND:5816
• NCBIGene:64689 (GORASP1 in the linked figure)
  • VS the result does have the chem D-glucose ("CHEBI:4167"), which has the equivalent ID PUBCHEM.COMPOUND:64689

I also see that in most cases, matchedCuries don't include the input Gene NCBIGene:154 (ADRB2)...but I think you said this was okay?
(My understanding is that the query process was adjusted - but PFOCR result-augmentation still runs on a template-level, where there are intermediate QNodes for intermediate genes to be bound to. So figures can match based ONLY on those intermediate genes)

[tokebe]

Fixed.