Factor graph-aggregated heterogeneous network embedding for disease-gene association prediction.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Understanding gene-disease relationships is crucial for disease pathogenesis research and therapeutic development.
• Computational methods accelerate the prediction of these associations.

Purpose of the Study:

• To propose FactorHNE, a factor graph-aggregated heterogeneous network embedding method.
• To address limitations of existing methods in utilizing multi-dimensional biological entity relationships from heterogeneous data.

Main Methods:

• FactorHNE captures semantic relationships between heterogeneous nodes via factorization.
• It generates distinct semantic factor graphs and aggregates diverse relationships.
• An end-to-end multi-perspective loss function optimizes the model for node embedding generation.

Main Results:

• FactorHNE effectively predicts disease-gene associations by leveraging node embeddings.
• The method demonstrates superior performance and scalability compared to existing models.
• Experimental analysis confirms the efficacy of FactorHNE.

Conclusions:

• FactorHNE offers improved performance and scalability for disease-gene association prediction.
• The model exhibits good interpretability.
• It is extendable to large-scale biomedical network data analysis.