Add "druggable" annotations as a new source ingest (HPA and or Pharos IDG)

[bazarkua]

I raised this in the DINGO meeting today. @mbrush suggested it is a good discussion topic and asked me to open a new issue here.

This came from an older request by Christine to support questions like "druggable genes that are in pathways associated with diseases." Related discussion is here:
RTXteam/RTX-KG2#12

Right now we do not have a clear plan for representing "druggable" information in the translator-ingests Tier 1 knowledge graph outputs. Without it, users cannot easily filter genes or proteins by druggability when doing pathway plus disease analysis.

Candidate knowledge sources mentioned so far:

@dkoslicki mentioned Human Protein Atlas has a "druggable proteome" page that looks like it could be used as a source for a druggable tag:
https://www.proteinatlas.org/humanproteome/tissue/druggable

@vdancik also mentioned Pharos as a druggability related source from the Illuminating the Druggable Genome program:
https://druggablegenome.net/
https://robokop.renci.org/api-docs/docs/automat/pharos

Questions for discussion:

Do we want to include a druggable tag in translator-ingests at all?

If yes, which source should we treat as the preferred source of truth?

What is the preferred modeling in KGX output. Should this be a boolean like druggable=true on gene or protein nodes. Or should this be modeled as explicit edges with a defined Biolink predicate?

[dkoslicki]

Relevant comment from @Mathelab : RTXteam/RTX-KG2#12 (comment)

Hi All,
Reminder that we do have the TDLs (Target Development Levels) updated from Pharos. Regarding 'druggable' definition, Tclin and Tchem could certainly be tagged as druggable. Here are the definitions again:

Tclin
These targets have activities in DrugCentral (ie. approved drugs) with known mechanism of action. In addition, updates from a curated list (https://doi.org/10.1038/d41573-024-00057-9) are added.

Tchem
These targets have activities in ChEMBL, DrugCentral, and Guide to Pharmacology that satisfy the following activity thresholds:

Activity Thresholds Activity values from DrugCentral and ChEMBL must be standardizable to -Log Molar units AND meet the the following target-family-specific cutoffs:
GPCRs: <= 100nM
Kinases: <= 30nM
Ion Channels: <= 10μM
Non-IDG Family Targets: <= 1μM
In some cases, targets have been manually migrated to Tchem by human curation based on small molecule activities from other sources.
The TDLs are maintained and we have acquired an updated curated list of newly approved drugs (https://doi.org/10.1038/d41573-024-00057-9). We are working to update this into the Pharos production website.

We can provide a downloadable dump file with all human uniprot protein ids with TDL label and annotations to why those TDLs are assigned.