A network medicine framework for multi-modal data integration in therapeutic target discovery.

Area of Science:

• Computational biology
• Translational medicine
• Genomics

Background:

• High costs and attrition rates in drug development necessitate improved therapeutic target discovery methods.
• Current strategies often rely on single-feature or single-modality approaches, limiting their effectiveness.

Purpose of the Study:

• To develop and validate a network medicine-based machine learning framework for systematic, disease-specific therapeutic target prioritization.
• To identify novel therapeutic candidates for clear cell renal cell carcinoma (ccRCC).

Main Methods:

• Integration of single-cell transcriptomics, bulk multi-omic profiles, genome-wide CRISPR screens, and protein-protein interaction networks.
• Application of a machine learning framework to prioritize disease-specific targets.
• In vitro validation of predicted therapeutic candidates in ccRCC models.

Main Results:

• The framework successfully identified known ccRCC targets.
• Five novel therapeutic candidates were predicted, including ENO2 and LRRK2.
• ENO2 inhibition demonstrated the most potent anti-tumor effect, followed by LRRK2, which has existing Parkinson's disease inhibitors.

Conclusions:

• The proposed machine learning framework offers a scalable and generalizable strategy for advancing therapeutic target discovery.
• This network medicine approach moves beyond traditional heuristics to a data-driven methodology applicable across various diseases.