Context-sensitive network-based disease genetics prediction and its implications in drug discovery

Area of Science:

• Computational biology
• Genetics
• Network science

Background:

• Disease-gene associations are crucial for understanding diseases.
• Existing disease networks often use pairwise similarities, neglecting contextual relationships.
• Context-Sensitive Networks (CSNs) offer a novel approach to model disease associations.

Purpose of the Study:

• To develop and evaluate a CSN-based phenome-driven approach for disease genetics prediction.
• To investigate the translational potential of predicted genes in drug discovery.
• To compare the performance of CSNs against traditional Similarity-Based disease Networks (SBNs).

Main Methods:

• Constructed two CSNs by directly linking diseases with associated phenotypes.
• Integrated CSNs with a genetic functional relationship network.
• Employed a network-based ranking algorithm for gene prediction and validated against SBNs.

Main Results:

• The CSN-based approach significantly improved gene prediction accuracy compared to SBNs (average rank from 12.6% to 8.8%).
• CSNs demonstrated a higher area under the receiver operating characteristic curve (0.91 vs. 0.87).
• Predicted genes for Parkinson's disease showed high relevance to its pathogenesis, identifying a potential drug target.

Conclusions:

• CSNs provide a superior method for disease genetics prediction over SBNs.
• The CSN framework effectively identifies potential therapeutic targets for diseases.
• This approach holds significant promise for advancing drug discovery and precision medicine.