CoKE

CoKE extracts and annotates drug–protein relationships from the COVID Open Research Dataset (CORD-19) to support identification of therapeutic candidates for SARS-CoV-2.

Key Features:

• Data Source: Uses the COVID Open Research Dataset (CORD-19) as the literature corpus for mining drug and protein mentions.
• Ontological Tagging and Entity Resolution: Applies SciBiteAI ontological tagging and resolves entities to UniProt for proteins and DrugBank for drugs.
• Relationship Identification Algorithm: Employs a custom algorithm that detects co-occurrences of protein and drug terms and assigns confidence scores to identified pairs.
• Processed Results: Has processed CORD-19 to identify approximately 3,000 drug–protein pairs involving 29 unique proteins and over 500 investigational, experimental, and approved drugs, including compounds in clinical trials.

Scientific Applications:

• Drug repurposing for SARS-CoV-2: Prioritizes drug–protein associations to support identification of candidate therapeutics against SARS-CoV-2.
• Literature-scale knowledge extraction: Condenses the CORD-19 corpus into curated drug–target relationships for downstream analysis and hypothesis generation.

Methodology:

Automated extraction from CORD-19 using SciBiteAI ontological tagging; mapping of extracted entities to UniProt and DrugBank identifiers; detection of co-occurrences between drug and protein terms via a custom algorithm that calculates confidence scores for each pair.