Monday, May 11, 2020

The Clinical Knowledge Graph -- Is this the missing downstream piece!??!?


This deserves FAR more time than I have right now to give any one single thing right this second, but if it is 1/10 what I think I'm looking at, it's amazing and you should check it out.
How bummed out do you get when you give someone a perfect dataset -- you know the one, the one where the QA runs before and after were perfect. The one where you didn't need to impute anything. The one where you didn't even need to apply normalization because your loading was spot on in every injection. You hand that over knowing that at a technical level you've done EVERYTHING as well as anyone in the world could do it.....

....and you get the blank stare from the collaborator.....and that "what do I do with this?"

And what do you do?
Your first thought is to say something like, "Wait. What? Aren't you the f****** biologist? Isn't this the disease you study? Go do f****** biology with it... don't you dare tell me you brought this here with no plan of what to do next!?! What do you actually do here?!? F*** right off and get out of my office, F***!"

But that isn't what you can say, because you weren't raised in West Virginia and you were taught better manners than that. Instead you point them to the 4 resources that probably make the most sense for their study of the 50 or so that you know probably could be used to help them, because -- well -- we really don't have centralized proteomics knowledge or even protein level data. It's spread everywhere.

Maybe the Clinical Knowledge Graph is a step toward fixing this. Maybe? I can hope until I can put some data into it!

1 comment:

  1. The dataset is huge and is not downloadable by neo4j communtity users. It could be split and made available as separate csv files. (I need only medicinal and indications data). CKG seems to be vast and a commendable work!!

    ReplyDelete