not urgent: ideal KL/AT adjustments

[colleenXu]

I did a review of current KL/AT annotations with Matt Brush (8/9). He pointed out several cases where things could be improved, but it would take more effort to do them. So I'm recording them in this issue to remember to do later if we continue using these pending BioThings APIs...

Matt Brush is aware that these would take a while to do/may not get done in phase 2 (up to Nov 2024) and is okay with this.

BioThings DISEASES

could be more fine-grained based on evidence_value. Would need jmespath processing due to reverses situation.
(previously noted here and here under "trouble assigning both values").

• knowledge: knowledge assertion, manual agent. Matt's notes: from curated sources like MedlinePlus, other KBs
• text-mined: not provided, text mining agent. Matt's notes: their own text mining tool
• experiments: statistical association, data analysis pipeline. Matt's notes: actual statistical associations from TIGA GWAS studies

MyGene GO annotations (from NCBI-Gene)

could be more fine-grained based on evidence codes (Matt gave this mapping: ARAGORN-GO-Evidence-KLAT-Mappings.xlsx). Would need jmespath processing due to reverses situation + indexing to work on the evidence fields...
(previously noted here under "trouble assigning both values").

MyChem FDA orphan data

could be more fine-grained predicates and KL/AT based on designated_status? Would need jmespath processing due to reverses situation
(previously noted here under "trouble assigning both values").

Could try adding x-bte annotation/BTE-processing to create an approved/off-label/etc. edge-attribute.