pubmedKB: an interactive web server for exploring biomedical entity relations in the biomedical literature

Area of Science:

• Bioinformatics
• Computational Biology
• Biomedical Informatics

Background:

• Genomic data is rapidly expanding, necessitating efficient methods for biomedical researchers to access and interpret scientific literature.
• Current PubMed searches lack sophisticated tools for retrieving complex relationships between biomedical concepts.
• A gap exists in search engines capable of extracting and visualizing semantic connections within biomedical publications.

Purpose of the Study:

• To develop pubmedKB, a web server for extracting and visualizing semantic relationships between variants, genes, diseases, and chemicals.
• To provide researchers with an intelligent literature mining tool for rapid knowledge discovery.
• To enhance the exploration of human genetic information within the vast PubMed database.

Main Methods:

• Utilized state-of-the-art natural language processing techniques to extract semantic relations from PubMed abstracts.
• Processed over 33 million PubMed abstracts to identify more than 2 million semantic relations.
• Developed a user-friendly interface featuring an interactive semantic graph for querying and exploring entity relationships.

Main Results:

• Successfully extracted over 2 million semantic relations between biomedical entities.
• Created pubmedKB, a comprehensive knowledge base of interconnected biomedical concepts.
• Implemented an interactive graph visualization and supporting sentence snippets for intuitive navigation.

Conclusions:

• pubmedKB offers a novel approach to biomedical literature exploration, enabling intelligent and rapid searching.
• The tool facilitates the discovery of complex relationships, providing valuable insights for researchers.
• pubmedKB is accessible via a web server, empowering domain experts to interrogate genomic information effectively.