BioTAGME: A comprehensive platform for biological knowledge network analysis

A Knowledge Graph containing logical or physical relationships among biological elements. 
This Network was built through BioTAGME, a system that combines TAGME, an entity-annotation framework based on Wikipedia corpus with a network-based inference methodology (i.e., DT-Hybrid).
BioTAGME exploits several Biological Ontologies as "ground truth" of significant bio-entities, such as DisGeNET, DrugBank, STRING and many more.    
We Deployed BioTAGME on PubMed, where the aim was on extracting biological entities and their relations from titles and abstracts.
Biological entities are the nodes of our graph, while edges are the relations between them.
Edges are of three categories:

Literature edges: interactions derived from publications.

 STRING: protein-protein associations stored in the STRING database.

BioTAGME: interactions predicted by our tool.

The network is released (BiotagmeNetwork.zip) in a neo4j compatible format. Such archive contains:

Edges.csv and Nodes.csv that contain the nodes and edges of our network, rispectively.

Name_Aliases.csv: contains the synonyms list of each biological entity.

Wiki_Titles.csv: contains the wikipedia pages title associated with the annotated entities.

wid1_wid2_pmid.csv: contains the associations between pairs of entities and pubmed id

BioIDs_WikiIDs.csv: contains the association between the biological entities annotated by BioTAGME, and the wikipedia pages that are associated.