DKG4RS

DKG4RS: Drug Knowledge Graph for Recommender Systems

Introduced and employed in:

https://github.com/javadtayebi/EKGDR

Introduction

DKG4RS, a comprehensive drug knowledge graph, constructed for knowledge graph based drug recommendation. Drug knowledge graph serves as side-information alongside disease-drug known interactions, to predict (score) potential interactions between diseases and drugs.

Features

		DKG4RS
Disease-Drug Interaction	#Diseases	1,229
	#Drugs	1,509
	#Interactions	6,657
Knowledge Graph	#Entities	32,672
	#Relations	28
	#Triplets	1,441,310

• disease_drug_interactions.txt

  • disease-drug interactions file
  • Extracted from repoDB.
  • Each row is a disease with its approved positive (observed/known) interactions with drugs.

• kg_final.txt

  • Knowledge Graph (KG) triplets file
  • Extracted from DrugBank(v5.1.9), MetaADEDB (drug side effects), and PubChem (PubChem substructure fingerprint for drugs).
  • Each row is a triplet in the form of (head entity, relation, tail entity).

• disease_list.txt

  • diseases dictionary file
  • Each row is a triplet in the form of (original ID | name | ID in DKG4RS).

• drug_list.txt

  • drugs dictionary file
  • Each row is a triplet in the form of (original ID | name | ID in DKG4RS).

• entity_list.txt

  • KG entities dictionary file
  • Each row is a triplet in the form of (original ID | name | ID in DKG4RS).

• relation_list.txt

  • relations dictionary file
  • Each row is a pair in the form of (original ID (name) | ID in DKG4RS).