RWDisEnh+: Enhancing disease-enhancer association prediction through multiplex-heterogeneous networks

Area of Science:

• Genomics
• Computational Biology
• Systems Biology

Background:

• Enhancers are regulatory DNA elements crucial for gene expression, but their links to human diseases are poorly understood.
• Millions of enhancers identified lack characterized disease associations, necessitating advanced computational tools.
• Previous work introduced RWDisEnh, a network-based method for predicting disease-enhancer associations using gene similarity.

Purpose of the Study:

• To enhance disease-enhancer association prediction by developing RWDisEnh+.
• To integrate sequence-based enhancer similarity into a multiplex-heterogeneous network framework.
• To improve the accuracy and scope of identifying disease-associated enhancers.

Main Methods:

• Developed RWDisEnh+, an enhanced network-based prediction method.
• Incorporated a sequence-based enhancer similarity network into a multiplex-heterogeneous framework.
• Utilized an extended random walk with restart (RWR) algorithm for information propagation across disease and enhancer layers.

Main Results:

• RWDisEnh+ achieved an average AUC of 0.874, outperforming RWDisEnh (AUC 0.819).
• Identified numerous evidence-supported disease-enhancer associations, including 10 enhancers for seven diseases (e.g., asthma, rheumatoid arthritis, type 2 diabetes).
• Predicted associations were enriched in immune, inflammatory, and metabolic pathways via GWAS and pathway analyses.

Conclusions:

• RWDisEnh+ offers a robust and effective framework for novel disease-enhancer association prediction.
• The method provides insights into enhancer-mediated gene regulation in complex diseases.
• Highlights the biological relevance of predicted enhancer-disease links in immune and metabolic functions.