Multi-domain knowledge graph embeddings for gene-disease association prediction.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Predicting gene-disease associations requires integrating diverse data sources and advanced computational methods.
• Knowledge graph embeddings offer a way to represent genes and diseases using ontological knowledge for machine learning.
• Current methods often overlook the benefits of using multiple, interconnected ontologies for complex biological predictions.

Purpose of the Study:

• To develop a novel approach for predicting gene-disease associations using knowledge graph embeddings.
• To leverage rich semantic representations from multiple, interconnected ontologies.

Main Methods:

• Utilized knowledge graph embeddings over multiple ontologies linked by logical definitions and mappings.
• Explored rich semantic representations for improved gene-disease prediction.

Main Results:

• The proposed method significantly improved gene-disease prediction accuracy.
• Different knowledge graph embedding techniques showed varying benefits from semantic richness.

Conclusions:

• Knowledge graph embeddings across interconnected biomedical ontologies show promise for gene-disease prediction.
• This approach can be extended to other ontologies and tasks requiring multi-perspective data analysis.
• Software and data are publicly available for further research.