search

ExposuresProvider / cam-pipeline

Data loading pipeline for CAM database
https://exposuresprovider.github.io/cam-pipeline/
MIT License
2 stars 4 forks source link

Figure out everything we can connect to the 28 genes of interest to the ICEES/ROBOKOP work #128

Open gaurav opened 4 months ago

gaurav commented 4 months ago

@karafecho has been working on getting interesting genes from ICEES/ROBOKOP and wants to know everything CAM-KP knows about the 28 genes of interest. That should be something we can query via Automat-CAM-KP with an appropriate TRAPI query, but if not we might seem some additional tooling to get at that information. If nothing else, this will be an useful exercise in making sure the models we expose via TRAPI are usable.

For example, on TLR2 (NCBIGene:7097), we contain information from CTD such as:

karafecho commented 4 months ago

This worked!

cam-kp-kg-TLR2-biological-process_response_1707935441713.json

karafecho commented 4 months ago

I have a question, @gaurav . If I don't care about TRAPI, is there an endpoint at which I can run queries that provide richer CAM assertions on, e.g., NCBIGene:7097? I can totally work with the Automat-CAM-KP output, and there's definitely some interesting stuff there, but I'm wondering if there's a way to tap into richer CAM assertions. Kind of like we do with ICEES, with icees-kg restricted to precomputed one-hop correlations and ICEES+ capable of dynamic queries and sophisticated statistical analyses.

gaurav commented 4 months ago

There isn't at the moment. We used to be able to run SPARQL queries directly against the Blazegraph backend, but since we moved to Automat we haven't been updating that. @balhoff Should we consider turning a Blazegraph instance back on with the latest kg.tsv file so that we can build alternate frontends to CAM-KP?

balhoff commented 4 months ago

@gaurav would the cypher endpoint provided by automat satisfy that?

karafecho commented 4 months ago

Note that my question is relevant to #116 on aux graphs.

gaurav commented 4 months ago

@gaurav would the cypher endpoint provided by automat satisfy that?

Oh yes, that works! I tried MATCH (s)-[p]-(o) WHERE s.id = 'NCBIGene:7097' RETURN s, p, o LIMIT 10 and ended up with a bunch of connections between TLR2 and other concepts. I'll see if I can come up with some sort of quick UI for this next week.