RTXteam
document types of edge scores that are currently in KG2
Currently we have:
- Chembl drug-to-target scores
- SEMMEDDB scores
are there any others?
Let's report back here.
I ran:
And got the following output:
This suggests that
- SEMMEDDB
- HL7
- MTH
- MEDLINEPLUS
- ChemBL Compound
all provide edge scores.
Thank you, @ericawood and @kvarforl. In all cases are the scores in the subject slot of the publication info object underneath the publications_info property?
Also, both IntAct and DisGeNET have edge scores but due to issue RTXteam/RTX-KG2#23, I didn't store them anywhere (since they're not tied to a particular PMID and it seems a bit better to wait until we get definitive guidance from the EPC WG than go back and potentially redo work later). Both ETLs pull them out into variables so that, when we're ready, we'll have them ready to go.
If the score is a chi-square score or a p-value, there is already an association slot for it, in the Biolink model:
https://github.com/biolink/biolink-model/blob/8b5ca35ad171f5837ae3b61d3df31e5c3a04b344/biolink-model.yaml#L4665
But, to get it into Neo4j, that would necessitate creating a new column in the edges TSV file.
Are we capturing all of these types of edge scores into KG2c? @sundareswarpullela and @ecwood can you please check when you have time? At this point, I am aiming to gather information. Then we can decide about if there are specific types of edge scores we might want to backfill into KG2c. Thank you.