Steps to process the Rare Disease KG

1) Create a base knowledge graph for rare diseases. We provide example scripts for building a base graph (i.e. PrimeKG) and adding any missing Orphanet edges. Please refer to build_graph.ipynb and add_orphanet_data_to_kg.py. We also include the output of the add_orphanet_data_to_kg.py for your reference, but none of the input data for build_graph.ipynb or add_orphanet_data_to_kg.py.

2) Process and finalize knowledge graph for rare diseases. The script prepare_graph.py adds triadic closures between phenotypes, diseases, and genes, extracts the largest connected component of the knowledge graph, splits the knowledge graph edges into train, validation, and test sets, and produces the required knowledge graph inputs for SHEPHERD.