EpiGraphDB: a database and data mining platform for health data science

Area of Science:

• Population health sciences
• Biomedical informatics
• Epidemiology

Background:

• Growing biomedical data necessitates integration, curation, and mining for research efficiency.
• Challenges in data integration lead to mis-inference and hinder reproducible research.

Purpose of the Study:

• To develop EpiGraphDB, a graph database and analytical platform for human population health data science.
• To demonstrate the platform's utility in addressing complex epidemiological and biomedical research questions.

Main Methods:

• Developed EpiGraphDB, a graph database integrating biomedical and epidemiological relationships.
• Created an analytical platform to support data science applications.
• Conducted three case studies showcasing platform capabilities.

Main Results:

• EpiGraphDB facilitates evaluation of pleiotropic relationships, reducing mis-inference in causal analysis.
• Protein-protein interaction data integration aids in identifying novel drug targets.
• Integration of Mendelian randomization with literature data enables evidence triangulation.

Conclusions:

• EpiGraphDB provides a valuable resource for population health data science.
• The platform supports efficient data integration, causal inference, and discovery of new biological insights.
• Open availability of the platform and code promotes reproducible research.