DrugQuest - a text mining workflow for drug association discovery

Area of Science:

• Bioinformatics
• Computational Biology
• Cheminformatics

Background:

• Increasing volume of biomedical literature and databases necessitates advanced data mining techniques.
• Existing text mining methods primarily focus on PubMed, neglecting chemical databases.
• There is a need for specialized tools to extract insights from chemical databases like DrugBank.

Purpose of the Study:

• To develop and apply a text mining approach for exploring drug associations within the DrugBank database.
• To identify and group drugs based on shared textual characteristics in their descriptions, indications, and mechanisms of action.
• To facilitate knowledge discovery and uncover potential new relationships between drugs.

Main Methods:

• Applied text mining techniques to DrugBank's text fields (Description, Indication, Pharmacodynamics, Mechanism of Action).
• Utilized Named Entity Recognition (NER) to identify biological entities (chemicals, proteins, genes, pathways, diseases).
• Employed the TextQuest algorithm for identifying significant biological terms and applied various clustering techniques to group drug records.

Main Results:

• Successfully clustered DrugBank records based on common textual terms, revealing underlying associations.
• Generated interactive visualizations including clustered chemicals and tag clouds of significant terms.
• Demonstrated the capability to identify potential new drug associations through text analysis.

Conclusions:

• DrugQuest is a valuable text mining tool for knowledge discovery in pharmacology.
• The tool effectively clusters DrugBank records to reveal novel drug associations.
• DrugQuest is freely accessible online for researchers to explore drug relationships.