Medical Graph Rag

Graphs

To store graphs used in this work an instance of Neo4j with apoc plugins enabled is needed. To build the necessary image run docker build -t <image_tag> . command in the root directory of the project which will use neo4j community image and enable the use of apoc plugins.

To run the necessary instances of neo4j a docker-compose can be used, with the provided docker-compose-example.yml as a basis. In order to use it rename the file to docker-compose.yml, change the <neo4j-tag> to appropriate tag and provide <path_to_import_directory> for hationet. After that run docker compose up -d

Hationet

1. To import hationet data to neo4j follow the instructions from https://github.com/nickzren/hetionet/blob/main/hetnet/csv/README.md
2. Then use embed_graph method Embedder class

UMLS

1. To import data from UMLS first request a license at https://www.nlm.nih.gov/research/umls/index.html
2. Download level 0 metathesaurus and place the META folder inside ./umls/data folder
3. Run the import_umls method