DGHNE: network enhancement-based method in identifying disease-causing genes through a heterogeneous biomedical...

Area of Science:

• Computational Biology
• Genetics
• Bioinformatics

Background:

• Identifying disease-causing genes is crucial for understanding disease mechanisms and developing treatments.
• Current methods for identifying disease genes lack sufficient accuracy and efficiency.
• Advanced computational approaches are needed to improve gene identification power.

Purpose of the Study:

• To propose and validate a novel computational method, DGHNE (Disease Gene Heterogeneous Network Enhancement), for identifying disease-causing genes.
• To enhance the accuracy and efficiency of disease gene identification using a heterogeneous biomedical network.
• To improve mechanistic understanding of disease etiology and facilitate clinical applications in disease prevention and treatment.

Main Methods:

• Constructed a disease-disease association network using phenotype similarity.
• Developed a heterogeneous biomedical network integrating disease-disease and gene-gene networks via disease-gene associations.
• Enhanced the heterogeneous network using network embedding with Gaussian random projection, followed by network propagation for candidate gene identification.

Main Results:

• DGHNE significantly outperformed five other methods in identifying disease-causing genes, achieving the highest area under the ROC and precision-recall curves.
• The method demonstrated superior precision and recall in both cross-validation and predicting novel disease-gene associations using the DisGeNet database.
• DGHNE successfully identified candidate causal genes for Parkinson's disease and diabetes mellitus, with predicted genes enriched in relevant biological pathways and terms.

Conclusions:

• DGHNE is a powerful and accurate computational tool for identifying disease-causing genes.
• The network enhancement approach significantly improves the identification of disease-gene relationships.
• The findings provide valuable insights into disease etiology and support the development of targeted therapeutic strategies.