Clojurians Log v2

A video update & demo on the state of UnifyBio & the RCRF Pattern Data Commons, and brief discussion on use of LLMs/agents/AI just at the edges of the system: https://youtu.be/tb00zvRvTN4 For those without context, the PDC stores molecular and clinical data on rare cancers and is built around Datomic as its primary store & system of record. Datasets are distributed over multiple Datomic databases plus a little bit of parquet on s3 aligned with the data for single cell scale data. The system is driven by a metamodel (annotations in the database about kinds & attributes) both for import and interop/adaptors. We have a public user role now, so anyone can sign up and access public data! rolling out to a few communities of ‘friends & family’ & partners first: https://data-commons.rcrf-dev.org We use a simple datalog in plain json parser (though the service and clients support transit & sometimes edn) to handle client library use in R, Python, and Julia. The Clojure API can be run in read-only trusted peer (internal at RCRF only, basically, for this system) or client mode. There are terms for playing around with data in the commons through UI and API. The tl;dr on them is you can’t commercialize the RCRF-provided datasets under the generic public use license. The full schema is viewable here: https://data-commons.rcrf-dev.org/schema/pattern/0.3.8/