Plover

Plover is a fully in-memory Python-based service for hosting/serving Biolink-compliant knowledge graphs as TRAPI Knowledge Provider APIs. It’s entirely Dockerized and doesn’t use any intermediary database.

In answering queries, Plover abides by all Translator Knowledge Provider reasoning requirements; it can also normalize the underlying graph and convert query node IDs to the proper equivalent identifiers for the given knowledge graph.

Multiple KPs can be run on the same Plover; each is exposed at its own endpoint (see here for more info).

There are two things needed to serve a KG via Plover: 1. Nodes and edges files for the graph (TSV or JSON Lines format), hosted at a publicly-accessible URL (more info here) 2. A Plover config file for the graph (more info here)

The steps to then run/deploy Plover differ slightly depending on what you’re trying to do: 1. Run Plover in a dev fashion (see here) 2. Deploy a new KP on an existing Plover (e.g., the Multiomics Plover) (see here) 3. Deploy Plover on a new instance (see here)

Plover automatically stands up the required TRAPI /query and /meta_knowledge_graph endpoints, as well as an endpoint that exposes SRI test triples for the KP (/sri_test_triples). It also provides a few other dev-oriented endpoints, described here.